Author: Woodenscale AI

Wirestock has raised a $23 million Series A to expand its creator-made multimodal training data business. The tough part in this market isn’t just collecting more files. It’s packaging legally cleared, human-made content into something model builders can train on without a rights mess later. Co-founder and CEO Mikayel Khachatryan launched Wirestock in 2018 with Ashot Mnatsakanyan, Vladimir Khoetsyan, and Hovhanness Kuloghlyan, initially helping creators sell across stock marketplaces before pivoting into AI data supply in 2023. Nava Ventures led the round. SBVP, Formula VC, and I2BF Ventures joined in.

What is Wirestock and how does it work?

At this point, Wirestock is basically a two-sided marketplace for creative AI training data. On one side, it recruits photographers, videographers, illustrators, designers, filmmakers, 3D artists, and now even music creators. On the other side, Wirestock sells AI labs ready-made visual datasets or custom content packages tailored to specific training goals. Its public dataset library already includes stock images and vectors. It also includes high-speed video, sports footage, dance videos, raw UGC clips, and long-form video. The broader platform has more than 50 million assets available.

For contributors, the workflow is pretty straightforward. A creator fills out a profile and applies for projects that match their skill set. They submit sample work, then get matched to paid assignments if the quality bar is high enough. Wirestock tracks project earnings in a dashboard and pays monthly. There’s a catch. Applicants must complete an unpaid test task first, and Wirestock uses both AI and human review to decide who gets accepted.

For buyers, the pitch is speed plus control. Wirestock offers off-the-shelf material from its existing library while also building custom multimodal datasets for AI teams. That matters because training a creative model often needs very specific combinations of format and labeling. Style, subject matter, and rights clearance matter too. Wirestock has had to retrain internal teams to annotate and label content in more detail. It then built sales and enterprise functions that can work with hyperscalers instead of just stock-photo customers.

What’s changed from the old Wirestock is the level of manual work it removes. In its early days, Wirestock helped creators upload content once and distribute it across multiple stock marketplaces while managing tasks like keywording, titles, and model releases. The AI version keeps that operational DNA, but turns it toward data procurement and dataset assembly instead of stock distribution.

Who founded Wirestock and why did it pivot?

The founding story

Wirestock began with a simple creator-tools idea. Khachatryan and his co-founders were already heavy users of stock content in other businesses, and after talking with contributors in 2018, they realized that selling visual work online involved too many repetitive steps. The company’s first product helped photographers, videographers, and illustrators license and monetize work across multiple marketplaces from one place.

That original version got real traction. By 2022, the platform had more than 100,000 photographers. But the bigger opportunity showed up once AI labs started looking for creative material that wasn’t just scraped from the public web. Wirestock pivoted in 2023 and let creators opt out of the new AI data supply business. Khachatryan said the company was transparent about the shift and that “the majority” of contributors chose to stay in.

Why the founders fit this market

Khachatryan’s path into this business isn’t the usual pure-AI founder story. In a 2021 interview, he said he came from a mathematics, statistics, and finance background, spent about a year in finance, then left because he wanted to build technology products instead. That matters here because Wirestock’s business sits in an awkward middle ground: part creator economy and part data operations. It’s also part enterprise sales. It needs someone who can think about workflow, not just model hype.

The broader founding team fit the problem for a different reason. Wirestock’s company history says the four founders were longtime friends obsessed with tech and spent a lot of time talking directly with artists before settling on the original marketplace-distribution idea. That early focus on creator pain helps explain why the company could later repurpose its supply base into an AI data network instead of starting from zero.

Traction after the pivot

This is where the story gets more interesting. Wirestock has more than 700,000 artists and designers signed up to complete creative tasks on the platform, and it currently supplies multimodal data to 6 of the largest foundation model makers, though it won’t name them. Khachatryan also said the business is running at a $40 million annual revenue run rate and has paid out $15 million to contributors so far. The company employs 60 people.

The way that revenue evolved is telling. Khachatryan said many of the first AI deals were just “off the shelf,” using existing library content, but the market shifted toward custom requests. That’s a stronger position than being a commodity stock archive. Custom data is harder to source. It’s harder to QA and usually harder to replace.

The Wirestock funding round

The new round is a $23 million Series A. Nava Ventures led it, and SBVP — the fund co-founded by Sheryl Sandberg — joined alongside Formula VC and I2BF Ventures. The raise brings Wirestock’s total capital raised to about $26 million, so this is the company’s first big institutional step-up rather than one more small extension.

The money is earmarked for research and engineering. Product hiring is part of the plan too. Wirestock is also building enterprise software so AI labs can collaborate on datasets, which hints at a product expansion beyond pure supply brokerage. Because the company wants more creative coverage in areas like 3D, audio, and music, some of this capital will go toward broadening the contributor base and the internal labeling stack.

Competition and positioning

Wirestock isn’t trying to beat Scale AI or Surge AI at everything. That would be a bad plan. Those companies operate at massive scale across many kinds of data work, and Scale in particular was valued at more than $29 billion after Meta’s June 13, 2025 investment for a 49% stake. Surge has also been discussed in the market at valuations ranging from $15 billion to $25 billion.

Wirestock’s angle is narrower and, frankly, more defensible. It focuses on creative modalities — image, video, illustration, design, gaming, 3D, and potentially audio — where rights, aesthetics, and contributor consent matter a lot. That makes its closer alternatives a weird mix: general data-labeling giants like Scale and Surge. There are also labor marketplaces like Mercor, licensed-data marketplaces like Human Native AI, and old-school stock libraries that were never built for AI training from day one. Human Native AI, for example, raised £2.8 million in seed funding for licensed AI data.

Why does this Wirestock funding round matter?

This round matters because Wirestock has already proved there’s money in the pivot. A $40 million run rate is not seed-stage experimentation. It suggests the company found a real buyer category fast, then found a second gear in custom projects once labs stopped wanting generic image dumps and started asking for more structured multimodal inputs.

It also matters for creators. A lot of AI-data businesses treat human contributors as interchangeable labor. Wirestock is pitching something a bit different: creative professionals as a renewable supply chain for premium training data. That doesn’t erase the labor questions around unpaid screening tasks or platform power. But it does create a clearer commercial path for photographers, illustrators, filmmakers, and designers who otherwise might have been cut out of the AI stack entirely.

It matters for investors too because the bet isn’t just on one dataset library. Martignetti’s thesis was that multimodal data won’t only be useful for generating prettier images or better video. He argued it will matter for models tackling “real-world tasks.” If that’s right, then a company that can continuously source, curate, and refine rights-cleared creative data starts to look less like a marketplace and more like infrastructure.

How big is the market for Wirestock’s AI data niche?

The macro numbers explain why startups like this are getting funded now. Grand View Research projects the global multimodal AI market will reach $10.89 billion by 2030, growing at a 36.8% CAGR from 2025 to 2030. Separate data from Technavio says the AI training dataset market is expected to expand by $9.12 billion from 2025 to 2030, with a 28.9% CAGR over that stretch.

Those figures line up with the broader shift inside AI development. Labs want models that can reason across text, images, video, and audio, which means they need richer and cleaner source material than the web usually provides. That pushes demand toward curated, licensed, multimodal datasets. Especially in creative categories where copyright risk and low-quality labeling can wreck model output.

Conclusion: what to watch next

Wirestock funding at this stage is less about a flashy round number and more about whether the company can turn a creator marketplace into durable AI data infrastructure. It already has supply, revenue, and some serious customers. The next thing to watch is whether its enterprise software and expansion into 3D, audio, and music make it stickier — or whether bigger data firms decide creative multimodal work is too valuable to leave to a specialist.

Read how Dhruva Space secured ₹105 crore in Project Garud grant support to build a standardised 500 kg-class satellite platform designed to simplify satellite manufacturing, launch integration, and orbit operations for telecom, earth observation, and national security missions.

FAQ

– What is the latest Wirestock funding round?  

 Wirestock raised a $23 million Series A announced on May 14, 2026. Nava Ventures led the round, and SBVP, Formula VC, and I2BF Ventures also participated, bringing the company’s total capital raised to about $26 million.

– How does Wirestock work for AI labs and creators?  

 Wirestock connects AI labs with a network of more than 700,000 creators who produce or license visual and other creative assets for model training. Labs can buy ready-made datasets or commission custom sets, while creators apply to projects, get reviewed for quality, track earnings in a dashboard, and receive monthly payouts.

– Who founded Wirestock?  

 Wirestock was started in 2018 by Mikayel Khachatryan, Ashot Mnatsakanyan, Vladimir Khoetsyan, and Hovhanness Kuloghlyan. Khachatryan has said he came from a math, statistics, and finance background before leaving finance to build technology products for creators.

– Is Wirestock a stock photo company or an AI data company?  

 It started as a stock-distribution and creator monetization tool, but since 2023 it has been operating as an AI data supplier focused on multimodal creative content. That puts it in a fast-growing corner of the market tied to licensed training datasets, visual AI training, and multimodal model development rather than traditional stock photography alone.

Dhruva Space builds satellites, ground stations, and launch support for customers that need one company to take a mission from hardware to orbit operations. That’s why Project Garud matters: the Hyderabad startup has secured ₹105 crore in grant support under the Centre’s Research, Development & Innovation Fund to build a standardised 500 kg-class satellite platform for telecom, earth observation, and national security workloads. India still relies heavily on imported systems or slower custom spacecraft builds for many of these missions, and Dhruva is betting that a production-ready bus can compress that cycle. Founded in 2012 by Sanjay Nekkanti, Chaitanya Dora Surapureddy, Abhay Egoor, and Krishna Teja Penamakuru, the company is trying to turn satellite manufacturing from a bespoke engineering exercise into a repeatable industrial business.

What is Project Garud and how will it work?

At a practical level, Project Garud is Dhruva Space’s push to build a standard satellite platform that customers can plug payloads into instead of starting from scratch each time. That’s already how Dhruva thinks about its broader stack: customers define payload and orbit requirements, the company matches those to a satellite bus and integrates the hardware. It then arranges launch access and runs operations through its ground segment and mission software. Its hosted-payload LEAP missions are built on the P-30 satellite bus, while its P-Nu microsat platform scales up to spacecraft of as much as 500 kg with low Earth orbit optimisation, deployable solar arrays, 3-axis stabilisation, and orbit manoeuvrability.

That matters because Dhruva isn’t selling just a box in orbit. It has built orbital deployers and launch-integration services. Its ground stack includes telemetry, tracking and command, payload data management, health monitoring, and automated control features such as orbital path prediction and radio control. In plain English: a customer doesn’t just get a satellite bus. It gets the operating layer needed to keep a mission alive after launch.

The before-and-after for a buyer is pretty stark. Before, a small or mid-size satellite mission often meant coordinating a spacecraft vendor, launch broker, and ground-station provider across different contracts and timelines. Mission-ops software sat on top of that. Dhruva’s model pulls those pieces into one workflow. Its launch offerings already span PSLV, SSLV, Falcon, Vikram, hosted payloads, and standalone missions, while its ground network covers 13 stations across 10 nations with 99% uptime. For constellation customers that care less about one heroic spacecraft and more about deploying dozens — or hundreds — on schedule, a standardised platform like Project Garud could help.

Who founded Dhruva Space and what has it built so far?

The founding story

Dhruva Space was founded in 2012 with a pretty blunt ambition: help privatise parts of India’s space sector long before that became fashionable. The founding team is still unusually intact for a deeptech company this old. Sanjay Nekkanti is CEO, Krishna Teja Penamakuru is COO, Abhay Egoor is CTO, and Chaitanya Dora Surapureddy is CFO. The company operates from Hyderabad and has grown into a full-stack space engineering business rather than a single-product startup.

Founder fit and operating history

The clearest publicly visible operating background is Penamakuru’s. Before Dhruva, he worked as a software engineer at Cisco and later led software at Savitri Aquamonk, where he worked on sensor-linked farm management tools. He studied computer science at BITS Pilani and Arizona State University. That helps explain why Dhruva’s business isn’t only about spacecraft hardware — it also has a serious systems and software layer.

The rest of the founding team’s fit comes through the way the company is organised. Egoor has long been the engineering face of the business, Nekkanti has handled commercial scaling, and Chaitanya Dora has stayed on the finance side through a stretch when most Indian spacetech companies were still fundraising on vision more than revenue. That division of labour matters a lot in hardware. Deeptech startups usually don’t fail because the tech is impossible. They fail because operations, finance, launch timing, and customer delivery don’t line up.

Traction, fundraising, and competition

Dhruva isn’t pre-product anymore. It launched Thybolt 1 and Thybolt 2 in 2022 — tiny amateur-radio nanosatellites weighing about 700 grams each — and the company has since moved into larger missions. In August 2025, it deployed LEAP-1, its first commercial satellite mission, aboard SpaceX’s reusable Falcon 9. The company now has more than 200 employees, runs out of a 28,000 sq ft Hyderabad facility, and is preparing a 280,000 sq ft manufacturing site in Shamshabad designed for spacecraft up to 500 kg.

The financing story is getting busier too. Months before this grant, Dhruva moved to raise ₹38.7 crore in an ongoing pre-Series B round in February 2026 from investors including IAN Alpha Fund, GVFL, Blue Ashva Capital, and Pradeep Sinha, after raising ₹51.76 crore in November 2025. The new ₹105 crore RDIF support matters for another reason: it’s non-dilutive capital for a hardware-heavy roadmap.

Competition is real, though. In February 2026, IN-SPACe selected Dhruva Space, Astrome Technologies, and Azista Industries to develop indigenous small satellite bus platforms. Each startup received ₹5 crore under that programme. Those are the closest direct peers in the bus-platform race. Then there are adjacent Indian private-space companies — Skyroot in launch and Pixxel in earth observation. GalaxEye focuses on imaging payloads. Dhruva’s edge is that it sits closer to the infrastructure layer: spacecraft buses and deployers. It also handles launch integration, ground systems, and hosted payload access in one stack. That’s a very different pitch from selling imagery or a launch slot.

Why does Project Garud matter for Dhruva Space now?

Because this isn’t just another cheque. It’s a vote for manufacturing scale.

Dhruva says Project Garud will help cut dependence on foreign satellite systems and support annual output of 500 to 600 satellites. If that target sounds ambitious, that’s because it is. But ambition is kind of the point here. A company can’t supply constellation customers with artisanal hardware.

There’s also a strategic layer that’s hard to ignore. The platform is aimed at telecom, earth observation, and national security use cases. Those are exactly the categories where countries want domestic capability, predictable supply, and some control over mission architecture. Dhruva’s CTO Abhay Egoor called Project Garud the “industrialisation of satellite manufacturing from India.” That framing fits. It’s less about one spacecraft and more about whether India can produce standard spacecraft buses at volume. He also said the longer-term roadmap could extend to MEO and GEO-class missions, which tells you Dhruva doesn’t want to stay boxed into only smaller LEO jobs.

And unlike an equity round, a grant of this size lets the company spend on R&D and production without immediately diluting ownership again. For a business building hardware, qualification processes, and manufacturing capacity, that’s a big deal.

How big is India’s spacetech market for satellite manufacturing?

India’s spacetech market is projected to reach $77 billion by 2030, and that’s the broad demand story behind Project Garud. This isn’t only about launch anymore. It’s about satellite manufacturing and downstream data services. It also includes defence applications, communications infrastructure, and the plumbing that supports all of it.

The policy backdrop has shifted fast too. The Department of Science & Technology launched the ₹1 lakh crore RDI scheme in July 2025 to back sectors such as deeptech, AI, robotics, space, biotech, climate tech, and the digital economy. The latest Dhruva grant was formalised under that scheme’s Enterprise Technology Evaluation process. That’s another sign the government wants private firms doing more of the heavy lifting in strategic tech.

The rest of the Indian market has been moving at the same time. Earlier this month, Skyroot Aerospace raised $60 million at a pre-money valuation of $1.1 billion and became India’s first spacetech unicorn. Around the same period, GalaxEye launched what it described as the world’s first OptoSAR satellite and established communication with it, while Pixxel partnered with Sarvam AI on orbital AI data centres. That doesn’t make them direct substitutes for Dhruva. But it shows a sector spreading beyond launch headlines into spacecraft, payloads, software, and orbital infrastructure.

What happens after this Project Garud bet?

The next thing to watch is execution speed.

Dhruva Space now has a clearer shot at building a standard 500 kg-class satellite platform at a moment when India wants more domestic space hardware and private customers want faster deployment cycles. If Project Garud works, Dhruva won’t just be another Indian spacetech startup with a good story — it could become a repeat supplier of satellite infrastructure. That’s a much tougher business to build. It’s also a much more defensible one.

Read how Anduril Industries raised a $5B Series H led by Thrive Capital and Andreessen Horowitz to build autonomous military hardware and software-defined defense systems designed to modernize how governments buy and deploy military technology.

FAQ

– What is the latest funding news around Dhruva Space?  

 Dhruva Space has secured ₹105 crore in grant support for Project Garud under the Centre’s Research, Development & Innovation Fund. This isn’t an equity round, so it gives the company capital for R&D and manufacturing without the same dilution pressure as a priced venture round.

– How does Project Garud actually help customers?  

 Project Garud is meant to become a standard 500 kg-class satellite platform that customers can use for telecom, earth observation, and national security missions. The appeal is speed and repeatability — instead of commissioning a one-off spacecraft every time, buyers can plug payloads into a more production-ready satellite bus and use Dhruva’s launch and ground-operations stack around it.

– Who are the founders of Dhruva Space?  

 Dhruva Space was founded in 2012 by Sanjay Nekkanti, Chaitanya Dora Surapureddy, Abhay Egoor, and Krishna Teja Penamakuru. The company still reflects that four-way split in leadership, with CEO, finance, operations, and engineering all staying anchored in the founding team.

– Is Dhruva Space a satellite maker or a broader spacetech company?  

 It’s broader than a satellite maker. Dhruva builds satellite platforms and deployers. It also offers launch-integration services, ground stations, and mission-operations software, which is why Project Garud fits into a much larger full-stack space engineering strategy.

Anduril builds autonomous military hardware and the software layer that ties those systems together. The new Anduril funding round is huge: $5 billion in Series H financing at a $61 billion valuation, led by returning investors Thrive Capital and Andreessen Horowitz. The pitch to customers is blunt — governments still buy too much custom defense gear too slowly, and Anduril wants to sell faster, more software-defined systems instead. Founded in 2017 by Brian Schimpf, Palmer Luckey, Trae Stephens, Matt Grimm, and Joe Chen, the company now looks less like a startup experiment and more like a private-sector prime contractor in the making.

That jump matters because it came less than a year after Anduril raised $2.5 billion at a $30.5 billion valuation. It also came after the company doubled revenue in 2025 to $2.2 billion. That helps explain why investors are still writing giant checks into a category that used to scare off most venture firms.

What is Anduril and how does it work?

At the center of Anduril’s stack is Lattice, the company’s autonomy and command software. In plain English, Lattice pulls in data from sensors, towers, drones, vehicles, and other systems. It turns that data into a live operational picture, then lets operators or connected agents act on it through planning, tasking, and command tools. The platform is built around 3 core software layers — entities, tasks, and objects. Operators can track what matters, push commands to connected systems, and store or share mission data across the network.

That workflow is what makes the company more than just a drone maker. Data can be pushed into Lattice from a new sensor feed or robot, or pulled back out for planning and tasking. Developers can use the platform to build user-facing apps for command and control and situational awareness. Mission planning too.

The manual work it removes is the ugly part of defense operations that rarely gets discussed outside procurement meetings: too many disconnected systems, too many screens, too much operator stitching in the middle. Lattice’s streaming tools keep a real-time view of entities in the system, while task APIs let connected agents receive and update mission status as work happens. That means less swivel-chair coordination and a cleaner path from “we detected something” to “we assigned something to deal with it.”

For customers, the before-and-after is the real sell. Before, a military buyer might be juggling separate vendors for sensing, tracking, command software, and the autonomous vehicle itself. After, Anduril is trying to package more of that stack together — not perfectly, and not always exclusively, but in a way that’s much closer to a finished product than a science project.

Who founded Anduril and why did they start it?

The founding story

Anduril launched in 2017 with Brian Schimpf, Palmer Luckey, Trae Stephens, Matt Grimm, and Joe Chen. The group came together around a specific thesis: Silicon Valley was good at building software fast, defense procurement was not, and autonomy would matter a lot more if it shipped as a real product instead of a years-long research program. Anduril’s identity still follows that bet.

Why these founders fit this market

Palmer Luckey brought the headline-making founder profile. He created Oculus VR and sold it to Facebook in 2014 for about $2 billion, which gave Anduril instant credibility on hardware ambition and product storytelling. Schimpf, Stephens, and Grimm brought a different kind of market fit — they were part of the Palantir orbit, which meant deep exposure to defense, intelligence, and government software problems before Anduril existed. Joe Chen added early Oculus hardware experience. That rounded out a team that understood both physical systems and high-stakes software.

Execution record, traction, and the new raise

The company is long past the “promising defense startup” stage. It’s a live business with deployed products, and its 2025 revenue hit $2.2 billion after doubling over the year. In recent weeks, it announced work tied to a space-based Golden Dome missile defense effort for the U.S., a contract win from the Dutch Ministry of Defence, and a U.S. Army deal for battle manager software built on Lattice to analyze data from joint missile defense systems.

The fundraising history tells the same story. This new Series H brought in $5 billion at a $61 billion valuation, led by Thrive Capital and Andreessen Horowitz. The prior round, closed less than a year earlier, was a $2.5 billion financing at a $30.5 billion valuation led by Founders Fund. At the time, the firm said its $1 billion check was the largest it had ever written. Altogether, Anduril has now raised more than $11 billion.

How Anduril stacks up against Shield AI, Hermeus, Helsing, and the old guard

Anduril isn’t alone anymore. Shield AI raised $1.5 billion in Series G financing at a $12.7 billion post-money valuation on March 26, 2026, while also adding $500 million in preferred equity financing. Hermeus closed a $350 million Series C on April 7, 2026, reaching a $1 billion post-money valuation as it pushed deeper into high-Mach unmanned aircraft. Helsing is close to another massive round in Europe.

But the categories aren’t identical. Shield AI leans heavily into autonomy software, especially Hivemind, while Hermeus is an aviation play with a very different product and customer rhythm. Anduril’s edge is that it tries to sell an integrated mix of hardware and software, while still behaving more like a product company than a traditional contractor. Lightspeed’s description of the business gets at the difference: Anduril identifies problems, privately funds R&D, and tries to sell finished products off the shelf in months instead of waiting years for a government spec to trickle down.

There’s a catch, though. The Department of Defense doesn’t look eager to crown one breakout startup and call it a day. The Air Force recently chose Shield AI software to work with Anduril’s Fury autonomous fighter jet rather than hand the full hardware-and-software stack to one company. That’s healthy competition. It also means Anduril’s future probably depends less on monopolizing programs and more on becoming too useful to leave out.

Why does this Anduril funding round matter?

A round this big changes what Anduril can reasonably chase.

It gives the company room to carry the balance-sheet weight that comes with giant defense programs, international expansion, and manufacturing-heavy product lines. Software margins are nice, but autonomous aircraft, missiles, subsea systems, and air defense products eat capital fast. Anduril now has a lot more freedom to fund that buildout without behaving like a cash-constrained startup.

It also changes how customers read the company. A defense buyer signing up for long-cycle programs wants to know the vendor will still be around, still shipping, and still supporting deployed systems years later. This raise helps answer that question better than any branding campaign could.

There’s a symbolic piece too. Schimpf wrote, “When we founded Anduril in 2017, defense was not a category that attracted significant venture investment. That has changed meaningfully over the last several years.” That quote lands because it’s true. This isn’t just another oversized venture round. It’s proof that a category once treated like niche hardware has become mainstream growth investing.

What does Anduril funding say about defense tech now?

The market backdrop is doing a lot of work here. Global military drone spending alone was estimated at $47.38 billion in 2025, and one forecast expects it to reach $77.27 billion by 2033. North America held more than 39% of the market in 2025, and the U.S. military drone segment dominated domestic share. That helps explain why autonomy, ISR, and counter-drone systems keep pulling in capital.

And the demand story isn’t subtle. Defense ministries want systems that can be fielded faster, updated in software, and operated with more autonomy than the old model allowed. Europe is spending harder on modernization and strategic autonomy. The U.S. is still the center of gravity. And companies that combine AI-driven software with deployable hardware are getting far more attention than they were a few years ago.

That doesn’t mean every defense startup becomes Anduril. A lot won’t. But the market is rewarding companies that can show real products, real contracts, and a faster path from prototype to field use.

Final take on Anduril funding

This Anduril funding round feels like a line in the sand. The company isn’t being priced like an interesting upstart anymore. It’s being priced like a serious defense platform builder with enough capital to push into bigger programs, tougher customers, and a wider global footprint. The next thing to watch isn’t whether Anduril can raise money again — it’s whether it can turn all that money into durable program wins without losing the speed that made it valuable in the first place.

Read how Origin Lab raised $8M to turn licensed video game worlds into structured AI training data for world models, robotics, and physical AI systems.

FAQ

– What is the latest Anduril funding round?  

 Anduril’s latest financing is a $5 billion Series H at a $61 billion valuation. Thrive Capital and Andreessen Horowitz led the round, and it came less than a year after a $2.5 billion raise that valued the company at $30.5 billion.

– How does Anduril’s product actually work?  

 Anduril combines autonomous defense hardware with Lattice, its software platform for command, control, planning, and tasking. Lattice ingests data from sensors and vehicles, creates a live operating picture, and helps operators or connected systems act on that data in real time.

– Who founded Anduril?  

 Anduril was founded in 2017 by Brian Schimpf, Palmer Luckey, Trae Stephens, Matt Grimm, and Joe Chen. Luckey previously founded Oculus, while several of the other founders brought Palantir experience, which gave the company an unusual mix of consumer hardware instincts and defense software credibility from day 1.

– Is Anduril a drone company or a defense software company?  

 It’s really both, and that’s the point. Anduril sells autonomous systems, including aircraft and other hardware, but its differentiation comes from Lattice — the software layer that connects sensors, vehicles, and operators into one system instead of a pile of separate tools.

Origin Lab is building a platform that turns licensed video game worlds into training data for AI systems that need to understand movement, physics, and 3D space. The startup has raised an $8 million seed round as demand grows for better data for world models and physical AI. That matters because labs working beyond text don’t have the neat, internet-scale data firehose that helped large language models take off. Origin Lab launched in 2026. It was started by Anne-Margot Rodde, Antoine Gargot, and Colin Carrier — a founding team pulling from AI, gaming, creator platforms, and data engineering.

What is Origin Lab and how does it work?

At a basic level, Origin Lab sits between game publishers and frontier AI labs. It licenses game-world content directly from rights holders. It captures that content through its own software pipelines, enriches it with structured metadata, and delivers custom datasets to model builders. So the buyer isn’t getting random gameplay clips off the internet. It’s getting a researcher-ready package built to spec.

That capture layer is the interesting part. Origin Lab records game worlds with engine-level access, pulling in synchronized video, keyboard and mouse input, camera telemetry, and depth information from the render pipeline. In practice, that means a lab can train on more than pixels. It can also learn from how a scene changes, where the camera moved, what inputs caused that movement, and what the environment state looked like at the time.

The platform also goes beyond raw capture. Origin Lab is building software for enrichment, QA, search, annotation, packaging, and delivery. Its datasets can include gameplay footage, 3D worlds, motion capture, and digital assets, all structured for multimodal training instead of dumped into a folder and left for the buyer to clean up later. That cuts out a lot of ugly manual work — licensing, formatting, and validation.

For customers, the before-and-after is pretty stark. Before, a lab might scrape public footage, negotiate one-off deals, or try to build synthetic environments from scratch. With Origin Lab, the pitch is simple: ask for a rights-cleared dataset with the signals you need, and get something source-controlled and usable for training from day 1. Rodde summed up the company’s thesis in one blunt line: “That data essentially lives in video games.”

Who founded Origin Lab and what makes them credible?

The founding story

Origin Lab came together around a pretty specific bet: AI is moving from language into environments, and the next bottleneck is no longer model architecture alone — it’s access to better world data. Gargot has said the company grew out of a shared ambition with Rodde to build a rights-cleared content platform for the AI era, with Carrier joining after that early thesis was already taking shape. The team’s view is that AI shouldn’t keep feeding on scraped content when richer, licensed environments already exist.

The company is based in California, with San Francisco listed as its headquarters. It’s still tiny — LinkedIn lists it in the 2-10 employee range. That makes the early commercial progress more notable than the org chart.

Founder fit

Rodde is the commercial and partnerships operator in the trio. She serves as co-CEO and chief commercial officer at Origin Lab, and she also comes from the gaming creator economy, where she built Creators Corp. That background fits the job here because Origin Lab isn’t just selling software. It has to convince rights holders that their assets can become a business, not a legal headache.

Gargot is the technical builder on the AI side. Origin Lab lists him as CTO, and his profile points to 10+ years in data, machine learning, and AI engineering. That matters because the company’s product isn’t a simple marketplace listing. It needs capture systems, multimodal structuring, and data pipelines that can work across different titles and hardware setups.

Carrier brings product and platform experience from the creator internet. He now serves as co-founder, co-CEO, and CPO at Origin Lab. Before this, he built Oooh, and his earlier career included years at Twitch — enough that former colleagues have described him as one of the people behind TwitchCon. He also holds patents around remixable video content and per-frame metadata capture. That feels unusually relevant for a company built around structured audiovisual data.

Early traction and fundraising details

For a company that only surfaced publicly in 2026, Origin Lab already has some real signals. It has exclusive partnerships with more than 20 game publishers covering more than 50 titles, and it’s already under contract with a leading frontier AI lab. That’s the kind of traction investors care about because it shows both sides of the marketplace moving at once.

Lightspeed Venture Partners led the $8 million seed round. SV Angel, Eniac, Seven Stars, and FPV joined in, along with angel checks from Twitch co-founder Kevin Lin and Cruise founder Kyle Vogt. The money is earmarked for capture and enrichment tech, publisher partnerships, and the engineering and research teams building dataset creation, QA, search, annotation, packaging, and delivery systems.

How Origin Lab compares with synthetic data rivals

Origin Lab’s closest competition doesn’t fit into one clean bucket. On one side, there are synthetic-data companies like Parallel Domain, Datagen, and Synthesis AI, which generate or simulate training data for computer vision and autonomy use cases. On the other, there’s the old-fashioned alternative: scrape footage from the web, hire people to clean it up, and hope the legal and quality issues don’t explode later.

What makes Origin Lab different is that it’s not promising to invent worlds from scratch. It’s packaging licensed worlds that already contain physics, player behavior, scene structure, and controllable interactions. That gives it a cleaner rights story than scraped footage and a different data profile than purely synthetic pipelines. Investors are backing that supply advantage — exclusive publisher relationships plus source-level capture — more than a vague “AI for gaming” pitch.

Why are investors backing Origin Lab now?

This round matters because it validates a pretty sharp thesis: the supplier layer in AI can become a big business when the biggest labs all hit the same bottleneck at once. Faraz Fatemi at Lightspeed pointed to companies like Scale AI as proof that data vendors can scale revenue fast when they become essential infrastructure. His read was even simpler than that — “the bottleneck for all of them is data.”

For Origin Lab, the cash should help turn a clever thesis into actual operating infrastructure. More capture tech means broader support across titles, more enrichment and QA means less bespoke wrangling for each buyer. More publisher relationships mean the company can become a real supply hub instead of a one-off broker doing custom projects behind the scenes.

There’s also a timing edge here. Labs working on world models, robotics, and multimodal systems need data that shows cause and effect, not just nice-looking frames. If Origin Lab becomes the place where they source that data legally and reliably, it could matter a lot more than its current size suggests. If it doesn’t, it risks ending up as a services business with fancy branding. That’s the tension to watch.

How big is the market around Origin Lab?

The synthetic data generation market is still small by big-tech standards, but it’s growing fast. Grand View Research puts the market at $218.4 million in 2023 and projects it to reach $1.79 billion by 2030, a 35.3% compound annual growth rate. And the fastest-growing segment in that report is image and video data — the exact area where Origin Lab is operating, even if its datasets are licensed and structured rather than fully synthetic.

The other trend is broader than synthetic data. AI labs are shifting from text-heavy systems to models that need to reason about environments, actions, and state changes. That’s why the conversation has moved toward world models, robotics, simulation, and multimodal training. Once that shift happens, flat internet content starts looking a lot less useful. Highly structured interactive data starts looking expensive — and valuable.

And there’s a legal undertone here too. Scraped training data has already created messy public blowback, including the December 2024 noise around early Sora outputs that appeared to echo video game and streamer footage. Origin Lab is basically selling the opposite of that mess: consent, provenance, and cleaner inputs. That won’t solve every training-data problem. But it’s a much more serious answer than “just scrape more.”

What should you watch next at Origin Lab?

Origin Lab has a smart pitch and a credible founding team. It also has the right enemy: bad, flat, legally murky training data. That’s a real problem, and the company is attacking it with something more concrete than most AI infrastructure startups manage in their first year.

The next test for Origin Lab is scale. Can it keep signing publishers, standardize messy game data across lots of titles, and become a repeat supplier to major labs instead of a niche broker for special projects? If it can, this seed round will look cheap in hindsight. If not, the idea will still be good — just smaller than the hype around physical AI makes it sound today.

Read how Mind Robotics raised over $1B to build AI-powered industrial robots designed for complex factory tasks that traditional automation still struggles to handle.

FAQ

– What funding did Origin Lab raise? 

 Origin Lab raised an $8 million seed round in May 2026. Lightspeed led the round, with participation from SV Angel, Eniac, Seven Stars, FPV, and angels including Kevin Lin and Kyle Vogt. That gives the company both venture backing and operators who know gaming and autonomy firsthand.

– How does Origin Lab turn video games into AI training data? 

 It licenses content directly from game publishers and then captures structured signals from the games themselves. That includes video, player inputs, camera telemetry, depth data, and metadata around scene composition and environment state. That gives AI labs more useful material than ordinary gameplay footage.

– Who are the founders of Origin Lab? 

 Origin Lab was started by Anne-Margot Rodde, Antoine Gargot, and Colin Carrier in 2026. Rodde brings partnerships and gaming-creator experience, Gargot comes from ML and AI engineering, and Carrier has a background in Twitch and creator-video products. The team spans both data infrastructure and content distribution.

– Is Origin Lab a synthetic data company? 

 Not exactly. It overlaps with the synthetic data market because it serves AI labs that need structured training data, but its core product is licensed, source-captured data from real game worlds rather than entirely generated scenes. That puts it in a different category from companies like Parallel Domain, Datagen, or Synthesis AI.

Mind Robotics builds AI-powered industrial robots for factory floors. Its latest funding news is big even by 2026 standards: the Rivian spinout has raised another $400 million just 2 months after landing $500 million, pushing total funding past $1 billion. The bet here is simple enough — factories still rely on people for a lot of messy, variable work that classic automation hasn’t handled well. RJ Scaringe, Rivian’s founder and CEO, created Mind Robotics in 2025 to go after that problem directly after deciding existing startups weren’t built for industrial work at real scale.

What is Mind Robotics and how does it work?

Mind Robotics is building intelligent robotics for industrial deployment, starting on the factory floor rather than in homes or research labs. Its system is full-stack: AI models and purpose-built robotic hardware. It also includes the deployment infrastructure needed to put those machines into real manufacturing environments. That matters because the jobs it’s targeting aren’t simple pick-and-place routines — they’re dexterous, variable, reasoning-heavy tasks that still break a lot of conventional automation.

The workflow is pretty clear. It starts with production-scale data from active Rivian manufacturing lines and uses that data to train and refine the models. Then it validates the robotic systems in the same kind of live industrial setting where they’re meant to operate. That’s the “data flywheel” pitch: build with a customer, not for a hypothetical customer.

Mind also isn’t pitching a one-trick machine. The company wants a platform that can generalize across core factory tasks and work safely alongside humans. That puts it somewhere between old-school fixed automation and the current wave of humanoid robot startups chasing general-purpose labor.

Scaringe has also been unusually blunt about the hardware philosophy. He has said Mind is leaning toward more traditional factory robot designs, not flashy humanoids, and joked that “doing cartwheels does not create value in manufacturing.” He’s also hinted Rivian’s custom robotics processor could eventually be useful to Mind. That suggests the startup may inherit more than just factory access from its parent.

Who founded Mind Robotics and what makes it credible?

The company started as an internal idea, then became a standalone bet

Mind Robotics began in 2025 under the internal name “Project Synapse.” Scaringe said he started it because he didn’t think the existing robotics field had the right mix of industrial know-how and supply-chain realism. It also lacked the model-training advantages needed to automate factory work at scale. In March, he described the effort as an attempt to build “robotics with human-like skills.” Ambitious wording, sure. But it explains why the company is chasing dexterity and adaptation instead of narrow repetitive automation.

Scaringe has actual manufacturing scar tissue

This isn’t a founder wandering in from consumer software. Rivian’s 2026 proxy says Scaringe founded Rivian in June 2009 and led the company through its major technical and manufacturing milestones, including vertically integrated technology platforms, multi-program manufacturing capabilities, and major partnerships. He holds a B.S. from Rensselaer Polytechnic Institute and both an M.S. and Ph.D. in mechanical engineering from MIT’s Sloan Automotive Laboratory.

That background is the whole case for founder-market fit. Mind Robotics is trying to automate work inside demanding production systems, and Scaringe has spent more than a decade building one. In his own words, he didn’t want Rivian’s future manufacturing dependency tied to robotics startups that hadn’t industrialized a product or built real supply chains.

There’s already a track record of spinning projects out

Mind isn’t the first time Scaringe has carved a new company out of Rivian’s internal work. He also helped create Also, the micromobility startup that spun out in 2025 and has raised more than $300 million to date. That doesn’t make every spinout a winner, obviously. But it does show he’s trying to turn internal technology programs into separate venture-backed businesses instead of burying them inside one corporate structure.

Early signals are promising, but still early

Mind Robotics is not a mature commercial robotics company yet. Its materials frame Rivian as the initial deployment partner, and Scaringe told The Wall Street Journal in March that Mind expects a large number of robots to be deployed by the end of 2026. So the company has real-world validation conditions and a live factory environment. The next proof point is execution.

Getting systems onto the floor, keeping them reliable, and showing they can earn their keep.

The fundraising is massive, and the cap table tells a story

Kleiner Perkins led the new $400 million round, with participation from the venture arms of Volkswagen and Salesforce. It came just 2 months after a $500 million Series A, and after a $115 million seed round led by Eclipse in late 2025. That sequence puts Mind above $1 billion in total funding and at a valuation north of $3 billion in barely half a year.

Competition is crowded — but Mind isn’t copying the pack

Mind enters a robotics market that already has heavyweight names. Apptronik closed a $403 million Series A to push its Apollo humanoid into industries including automotive, electronics manufacturing, and logistics. TechCrunch has also grouped Apptronik with Boston Dynamics, Agility Robotics, and Figure as prominent names in the humanoid race. Mind’s positioning is different: fewer sci-fi demos, more task-specific industrial systems trained in active factories.

The legacy alternative is even less glamorous. A lot of factories still buy fixed-function robotics from established vendors and then bolt on integrators to make the systems usable. In warehouse robotics alone, major incumbents include ABB, Honeywell, KUKA, OMRON, and Yaskawa. Mind’s argument is that the next step isn’t just another robot arm. It’s a system that can handle variability without requiring a factory to be redesigned around the machine.

Why does Mind Robotics funding matter now?

This round matters because it turns Mind from an intriguing spinout into a company that can afford industrial speed. Building robots for real factories is brutally expensive — hardware, safety, deployment, model training, and long validation cycles all eat cash. A startup that wants to move from prototype to plant floor needs more than a flashy demo and a small seed check.

The investor mix matters too. Volkswagen’s venture arm isn’t just a financial name on the cap table; Volkswagen is already tied to Rivian through a software joint venture launched in November 2024, with plans to use Rivian’s architecture in future vehicles and up to $5.8 billion committed by 2027. Salesforce Ventures, meanwhile, gives the round a software-and-enterprise stamp that’s different from a pure robotics wager.

For Rivian, there’s a deeper implication. If Mind works, Scaringe may have found a way to turn one automaker’s manufacturing pain into a standalone robotics business — using real factory data as the moat instead of treating operations as a cost center. That’s a much more serious thesis than “humanoids are cool.”

What does Mind Robotics funding say about factory automation?

The timing isn’t random. The International Federation of Robotics says 4.664 million industrial robots were operating worldwide in 2024, up 9% year over year, and annual installations in 2024 were the second-highest ever recorded. In the Americas alone, installations stayed above 50,000 for a fourth straight year, with the U.S. accounting for 34,200 units. This is already a huge operating market, not a science experiment.

There’s also room above the classic automotive-robot model. Warehouse robotics was a $4.31 billion market in 2022 and is projected to reach $17.29 billion by 2030, with a 19.6% CAGR. Growth is being pushed by labor shortages and e-commerce pressure. Safety needs also matter, as does the need to raise throughput without endless hiring. That’s the kind of environment where investors start paying up for more adaptable automation.

So the macro read is this: factories don’t just want more robots. They want robots that can deal with variation and work near people. They also need to fit into existing operations without months of custom engineering every time something changes. That’s the opening Mind is trying to exploit.

Conclusion

Mind Robotics funding is starting to look less like a side project and more like a serious industrial platform bet. The money is there. The founder credibility is there. The missing piece is the hard one — proving that these robots can move from promise to uptime inside Rivian plants by the end of 2026, and then beyond them.

Read how upGrad raised ₹360 Cr to expand AI-led learning, strengthen workforce training, and support its proposed Unacademy acquisition.

FAQ

– What funding did Mind Robotics just raise?  

 Mind Robotics just raised $400 million in a new round led by Kleiner Perkins. That came only 2 months after a $500 million Series A and follows a $115 million seed round from late 2025, pushing total funding above $1 billion and valuing the company at more than $3 billion.

– How does Mind Robotics work?  

 Mind Robotics combines AI models and purpose-built factory robotics hardware. It also includes deployment infrastructure in one industrial automation stack. The key twist is that it trains and refines those systems using production-scale data from Rivian’s active manufacturing lines, then aims to deploy the robots in those same environments.

– Who founded Mind Robotics?  

 RJ Scaringe created Mind Robotics in 2025 and serves as its chairman. He’s the founder and CEO of Rivian, which he started in June 2009, and he has degrees in mechanical engineering from RPI and MIT — a background that makes a lot more sense for industrial robotics than a generic software résumé would.

– Is Mind Robotics a humanoid robot company?  

 Not really — at least not by the way Scaringe has described it so far. Mind is focusing on industrial robots for factories and has emphasized more traditional or purpose-built designs over humanoid showpieces, putting it in the broader factory automation and industrial robotics category rather than the pure humanoid race.

upGrad is an online higher education and workforce training company, and its latest upGrad funding round brings in ₹360 crore as it lines up the proposed Unacademy acquisition and a bigger push into AI-led learning. The problem it’s chasing is pretty simple: Indian learners and employers don’t want one-off courses anymore — they want outcomes, credentials, mobility, and often a job link at the end. Founded in 2015 by Ronnie Screwvala, Mayank Kumar, Phalgun Kompalli, and Ravijot Chugh, upGrad is now trying to turn itself into a broader learning-and-work platform instead of staying just another course marketplace. That’s why this internal round matters more than the headline number suggests.

What is upGrad and how does it work?

upGrad sells structured learning, not random video libraries. A learner typically comes in through a goal — promotion, certification, AI training, a degree, study abroad, or career transition — and then moves through a guided program with coursework, mentorship, projects, support, and in many cases career services. Its homepage now spans online programs, offline learning options, study abroad, and newer AI-focused courses. That tells you how wide the company wants the funnel to be.

One of the clearest examples is upGrad Abroad. The model lets students begin online in India and then move on-campus later, with credits accepted at 18 global universities on selected pathways. The offering also bundles counselling, application support, visa help, and course selection. It’s the messy admin work that usually scares off first-time international applicants. And yes, the pitch is cost and flexibility: start at home, spend less, then finish overseas.

There’s a separate enterprise layer too. upGrad for Business works with employers on workforce skilling programs, especially around digital, leadership, and tech capabilities. That’s different from selling a single course to a consumer. It turns the company into a training vendor for firms that need custom learning tracks for teams. A steadier business, if it works.

What upGrad removes, basically, is fragmentation. Before platforms like this, a learner might juggle one provider for exam prep, another for certifications, a consultant for overseas admissions, and a separate route for placement help. upGrad’s bet is that people will pay for one guided system if it cuts confusion and gets them from “I need a better career move” to “here’s the credential, pathway, and support.”

Who founded upGrad and why are investors still backing it?

The founding story

upGrad started in 2015 after the founding team zeroed in on a blunt gap: professionals tend to stop structured learning once they enter the workforce, even as industries keep changing under them. The company was built around that tension from day one — not school tutoring, but working-professional upskilling tied to career progression. Ronnie Screwvala is co-founder and chairman, Mayank Kumar is co-founder and managing director, and Phalgun Kompalli is co-founder. Ravijot Chugh is also a co-founder and product leader.

Why the founders had market fit

Screwvala wasn’t coming in as a first-time operator. Before upGrad, he built UTV into a major media company and exited it to Disney in 2012 at an enterprise value of about $1.4 billion. That matters because edtech at scale isn’t just about course content — it’s a consumer business and a brand business. It’s also an execution business. upGrad’s original setup paired that scale-building experience with a team focused on product, curriculum, and learner outcomes for professionals.

Traction, product spread, and where the company sits now

This isn’t a pre-launch story. upGrad is already live across consumer courses, offline centres, study abroad, and enterprise skilling. The company has crossed 10 million enrolments across 100+ nations. That gives some sense of the reach it built before this latest round.

Fundraising details and the Unacademy angle

The fresh capital is internal, which stands out on its own. Ronnie Screwvala put in ₹300 crore, while existing investors Temasek, IFC, and 360 One joined on a pro-rata basis; Temasek contributed ₹45 crore, and IFC plus 360 One added around ₹16 crore together. The round totals ₹360 crore and comes just as upGrad prepares to close its proposed Unacademy acquisition. Days before this financing surfaced, upGrad filed a merger application with the Competition Commission of India on May 7, 2026.

The use of funds is broad, not narrow. upGrad plans to spend across test prep and study abroad. Enterprise skilling, workforce development, and AI-led learning products are also on the list. Moneycontrol first reported in November that the two companies were discussing a deal valued at $300 million to $400 million, and those talks were on-and-off for months before resuming and moving toward a term sheet this year. In its CCI filing, upGrad said the Unacademy buy would give it entry into online test preparation, where it doesn’t currently operate.

Who upGrad is competing with

upGrad’s real competition depends on the vertical. In professional upskilling, it runs into players like Simplilearn, Great Learning, Coursera, Udemy, and Emeritus. In study abroad, it competes with counsellors, pathway providers, and the old-school education-consultancy model. If the Unacademy deal closes, the battleground gets tougher — UPSC, JEE, NEET, and GATE are already crowded with established brands and offline coaching muscle. upGrad’s edge is that it’s trying to bundle degrees and skilling. Mobility and now potentially test prep sit in the same company. That’s ambitious. It’s also messy to execute.

Why does this upGrad funding round matter?

Because this isn’t just growth capital. It’s control capital.

An internal round led so heavily by the founder usually signals speed and conviction more than optics. upGrad didn’t bring in a flashy new outside lead; it leaned on insiders who already know the company and still wanted to write checks. Ronnie Screwvala alone putting in ₹300 crore tells you this wasn’t treated like a small bridge or a token top-up. It was meant to give the company room to make a strategic move.

That move is Unacademy. And if the deal closes, upGrad expects to have nearly ₹900 crore in cash tied to that transaction, taking its overall capital pool to roughly ₹1,260 crore. That gives it ammunition to go after adjacent categories instead of staying boxed into working-professional education. It also lets the company spend on offline and online formats at the same time. Expensive, but probably necessary now that pure digital edtech has lost some shine.

There’s another reason this matters. upGrad is trying to be an integrated learning-and-work platform at a moment when standalone edtech models look thin. Test prep, study abroad, enterprise skilling, certifications, placements, and AI products don’t naturally fit together unless the customer journey is designed really well. If upGrad can stitch those pieces together, the round looks smart. If it can’t, this becomes a holding pattern dressed up as expansion.

How big is India’s edtech market right now?

The market is still large, even after the post-pandemic reset. Grand View Research estimates India’s education technology market generated $6.6 billion in revenue in 2024 and could reach about $17.0 billion by 2030, growing at a 16.9% CAGR from 2025 to 2030. It also calls India the fastest-growing regional market in Asia Pacific in this category.

The shape of demand has changed too. Investors and customers now care a lot more about job-linked learning, blended delivery, and business models that don’t depend on endless discounting. That’s part of why Indian edtech companies have been broadening into enterprise skilling, overseas education, and hybrid formats instead of betting everything on one audience. The sector is still in consolidation mode, but it’s a more disciplined version than the 2021 boom years.

Final take on the upGrad funding round

The upGrad funding round is really a bet on consolidation.

₹360 crore by itself is useful, sure. But the bigger story is that upGrad wants to own more of the learner lifecycle — from test prep and degrees to study abroad and workforce training — while layering AI into the product mix. The next thing to watch is whether the Unacademy deal closes on schedule, and whether upGrad can turn a collection of adjacencies into one business that actually feels coherent.

Read how Exaforce raised $125M to build an AI-native security operations platform that helps enterprises detect, investigate, and respond to cyber threats with autonomous AI agents.

FAQ

– What is the latest upGrad funding round? 

 The latest upGrad funding round is an internal financing of ₹360 crore. Ronnie Screwvala led it with ₹300 crore, while Temasek, IFC, and 360 One also participated, as the company prepares to close its proposed Unacademy acquisition.

– How does upGrad actually work for learners? 

 upGrad works like a guided learning platform rather than a loose course catalog. It offers structured programs in higher education, skilling, study abroad, and enterprise learning, and for some study-abroad pathways it lets students start online in India before moving to partner universities overseas.

– Who founded upGrad? 

 upGrad was founded in 2015 by Ronnie Screwvala, Mayank Kumar, Phalgun Kompalli, and Ravijot Chugh. Screwvala brought major company-building experience from UTV, while the broader founding team built the platform around working professionals who needed to upskill without leaving their jobs.

– Is upGrad a test-prep company or an upskilling company? 

 Right now, it’s primarily an upskilling, higher education, study-abroad, and workforce-training company. But if the Unacademy acquisition goes through, upGrad will gain a direct route into online test prep categories like UPSC, JEE, NEET, and GATE, which would make it a much broader edtech player than it is today.

Exaforce builds software that helps security teams detect, investigate, and respond to cyber threats with AI agents instead of bouncing between a stack of separate tools. The Bay Area startup has now raised $125 million in fresh funding to push deeper into agentic security operations, a category getting crowded fast as enterprises look for faster answers to AI-assisted attacks. The pitch is simple: defenders can’t keep up when alerts pile up and zero-days spread quickly. Analysts still spend too much time reconstructing context by hand. Exaforce was founded in 2023 by Ankur Singla, Ariful Huq, Jakub Pavlik, and Devesh Mittal.

What does Exaforce’s agentic security operations platform do?

Exaforce is building an AI-native SOC platform that collects cloud and SaaS telemetry into a real-time security knowledge graph. The platform uses AI agents for threat detection, triage, investigation, and response through tools like Exabot Detect, Exabot Triage, Exabot Investigate, and Exabot Respond. It’s meant to handle detection and triage. Investigation and response happen inside one operating layer.

What makes that interesting is the order of operations. Many AI security tools still rebuild context during investigations by checking logs, APIs, and tickets one step at a time. Exaforce does the harder work upfront by connecting identities, permissions, configurations, files, code, and cloud activity as data arrives. That lets its agents retrieve context instead of rebuilding it on demand. In the company’s words, that cuts typical investigations from many minutes to under a minute.

The tooling looks a lot more like an analyst workspace than a chatbot wrapper. Exaforce’s platform supports natural-language queries and a conversational, visual data explorer. The platform also includes built-in workflows and a data pipeline that organizes and connects large volumes of telemetry data efficiently. This helps security teams avoid switching between SQL queries, scripts, APIs, and multiple dashboards during investigations.

There’s also a technical bet underneath all of this. Exaforce says its multi-model AI engine combines semantic, behavioral, and knowledge models rather than leaning on an LLM alone. The idea is to reduce hallucinations and improve consistency. It also makes the output more auditable for security teams that can’t afford fuzzy answers. The startup has also introduced “vibe hunting,” a natural-language approach to threat hunting that replaces rigid query-based searches.

Who founded Exaforce and how is it positioned in agentic security operations?

How Exaforce started

Exaforce came out of a pretty specific frustration. The founding team had experience securing large cloud environments and saw SOC teams struggling to stay ahead of threats. Other team members had worked on large language models and believed security teams needed specialized AI systems instead of generic models. They started Exaforce to speed up complex security tasks with specialized AI agents and better data infrastructure.

Why this team fits the problem

Ankur Singla gives the startup real founder-market fit. Before Exaforce, he founded Contrail Systems, which Juniper Networks acquired, and later founded Volterra, which F5 acquired. That history matters because Exaforce isn’t selling a lightweight AI add-on. Exaforce is building a deep infrastructure platform for enterprise security teams, and Singla has experience building and selling enterprise software companies.

The broader team adds credibility too. Some early Exaforce team members came from F5 and Palo Alto Networks, while others had worked on LLMs at Google. Co-founder Jakub Pavlik previously co-founded tcp cloud, later acquired by Mirantis, and held engineering leadership roles at Volterra and F5. The team brings experience across cloud infrastructure, security operations, and AI engineering.

Traction, fundraising, and competition

The startup is no longer in pure demo mode. Exaforce brought its product to market in the fourth quarter of 2025 after roughly 2 years of work with design partners. Since launch, it has added 20 customers, including Replit and Guardant Health, and expects to reach 40 to 50 customers by the end of 2026. Mayfield also said the company has grown to more than 130 employees and processed millions of investigations.

On the money side, Exaforce has raised a $125 million Series B after a $75 million Series A last year, bringing total funding to $200 million. The new round included HarbourVest, Peak XV, Mayfield, Khosla Ventures, Seligman Ventures, and AICONIC. TechCrunch reported the Series B valued the company at $725 million. Exaforce says the money will go into its core platform, especially the multi-model AI stack and the real-time knowledge graph. It also plans expansion in Japan and Europe, along with more spending on customer success, research, MDR oversight, and support.

Competition is already intense. Startups such as 7AI, Dropzone AI, and Prophet Security are chasing the same AI-for-SOC budget, while bigger platforms from Palo Alto Networks and CrowdStrike still own a lot of buyer attention. Exaforce’s clearest point of separation is architectural: it argues that legacy SIEMs and many AI SOC tools still reconstruct context mid-investigation, while its system keeps live context ready at ingest. That’s the bet investors are backing — not just more automation, but a faster and cheaper way to reason about security events at enterprise scale.

Ankur Singla put that positioning plainly: “We built Exaforce to be the platform defenders actually work in, not just an AI layer on top of existing tools.” Vinod Khosla framed the investor case just as directly: “When the cost of defence drops by an order of magnitude, the entire calculus of security changes.”

Why did Exaforce raise $125M now?

This round matters because Exaforce is trying to cross a hard boundary: from impressive early product to global security platform. That jump gets expensive fast. Selling into enterprise security teams means long deployments, high support expectations, constant product tuning, and a real services layer around the software. Exaforce isn’t just hiring researchers. It’s putting money into MDR oversight, customer success, and support.

The timing also says something about investor appetite. A year after the Series A, the company came back with a much larger round and one of the bigger financings in the emerging AI SOC segment. That usually means two things: the product is landing with customers, and investors think the window to establish category leadership is open right now, not 3 years from now. Mayfield’s note about production deployments, millions of investigations, and a tripled valuation backs that up.

For customers, the practical implication is simple. Exaforce now has the cash to expand product breadth and geographic reach without slowing down. If the company executes, buyers should expect a more mature platform and tighter workflows across detection and response. They should also expect stronger local coverage in markets beyond the US.

How big is the market for agentic security operations?

The market Exaforce is chasing is large enough to attract a lot of capital and messy enough to create room for new platforms. Frost & Sullivan estimates the global modern SIEM market will grow from $7.13 billion in 2024 to $13.55 billion by 2029, with cloud SIEM revenue growing at a 17.5% CAGR over that period. That’s a useful proxy because modern SIEM platforms are steadily absorbing analytics, automation, and AI-assisted investigation features that used to sit in separate tools.

The services side is moving too. Grand View Research estimated the managed detection and response market at $3.5 billion in 2023 and projects it will reach $15.31 billion by 2030. That lines up with what buyers are actually asking for now: fewer point tools and faster triage. They also want a mix of software plus expert oversight when in-house teams are stretched thin.

And that’s why Exaforce’s timing doesn’t feel random. AI-driven attacks are pushing enterprises to modernize security operations, but the real pressure is operational. Security teams need automation that’s explainable, fast, and cheap enough to run continuously across cloud-heavy environments. The winners probably won’t be the noisiest AI vendors. They’ll be the ones that can fit into day-to-day SOC work without making analysts trust magic.

What should customers watch next from Exaforce?

Exaforce has enough money now to stop being judged like an intriguing startup and start being judged like a platform company.

That changes the standard.

The next thing to watch is whether its agentic security operations story holds up under broader deployment. Can it expand beyond early believers, keep false positives low, and make the human-plus-agent workflow feel reliable enough for real incident response? If it can, this round will look smart. If it can’t, it’ll be another reminder that AI security products sound a lot better in pitch decks than they do at 3 a.m. in a live SOC.

Read how Nivasa Finance raised ₹25 Cr in seed funding from Prime Venture Partners, Blume Ventures, and Whiteboard Capital to simplify affordable home-loan access for rural and semi-urban borrowers across India.

FAQ

– What is Exaforce’s latest funding round?  

 Exaforce raised a $125 million Series B in May 2026. The round followed a $75 million Series A a year earlier and pushed total funding to $200 million, with investors including HarbourVest, Peak XV, Mayfield, Khosla Ventures, Seligman Ventures, and AICONIC.

– How does Exaforce work for security teams?  

 Exaforce gives SOC teams a single platform that unifies telemetry, builds a real-time knowledge graph, and then uses AI agents to detect, triage, investigate, and respond to threats. The product includes Exabot Detect, Triage, Investigate, and Respond, plus natural-language search and visual investigation tools that reduce the need for manual queries and scripting.

– Who founded Exaforce?  

 Exaforce was founded in 2023 by Ankur Singla, Ariful Huq, Jakub Pavlik, and Devesh Mittal. Singla previously built Contrail Systems and Volterra, while Pavlik had already co-founded tcp cloud and later led engineering work at Volterra and F5, which gives the team a strong mix of cloud, infrastructure, and security depth.

– Is Exaforce a SIEM company or an MDR company?  

 It sits somewhere in between, which is part of the appeal. Exaforce is pitching an AI-native security operations platform that overlaps with modern SIEM, SOC automation, and MDR by combining data ingestion, investigation, response workflows, and optional managed oversight in one system.

Nivasa Finance is a Bengaluru startup that connects underserved home-loan borrowers with banks and NBFCs for affordable housing finance. The company has raised ₹25 crore in seed funding from Prime Venture Partners, Blume Ventures, Whiteboard Capital, and several angel investors. Nivasa Finance wants to help rural and semi-urban borrowers access formal home loans more easily. Founded in 2025 by Samit Shetty and Hitesh Saraf, the startup is building a faster and clearer housing loan process. Easier to execute on the ground, too.

What is Nivasa Finance and how does it work?

Nivasa Finance isn’t a bank. It works as a distribution and fulfilment layer for secured housing credit, helping borrowers with onboarding and documentation. It also matches them with lenders and handles loan disbursal, while the actual loan sits with partner banks, NBFCs, small finance banks, or housing finance companies. That matters because the company is trying to remove the part of the journey where customers bounce between branches, agents, and paperwork without knowing which lender will actually say yes.

The product flow is pretty direct. A borrower or local advisor can start a lead through WhatsApp or Nivasa’s online portal, submit personal and financial details digitally, and move into remote assessment before the case is routed to the most suitable lending partner. Nivasa has also built customer-facing digital interfaces and app-based journeys. It still keeps doorstep service in the loop for people who need hand-holding offline.

That hybrid model is the point. Nivasa doesn’t pretend affordable housing borrowers in non-metro India will all self-serve through an app. Instead, it combines field advisors and telecalling with screening and branch coordination through digital workflows, so the lender gets cleaner files and the borrower doesn’t have to decode the mortgage process alone. Its Mysuru branch, opened in June 2025, was built to support walk-ins and advisor engagement. It also handles lead screening, telecalls, and disbursal coordination.

The service is pitched around small-ticket home loans — the website advertises loans from ₹5 lakh to ₹35 lakh — and it doesn’t charge customers commission or service fees. On the partner side, Nivasa lists registered relationships with lenders such as Slice Small Finance Bank, Muthoot Housing Finance, Jana Small Finance Bank, and Veritas Finance. That gives a clearer picture of the kind of institutions it’s plugging into.

Who founded Nivasa Finance and what makes them credible?

The founding story

Nivasa Finance was founded in 2025 by Samit Shetty and Hitesh Saraf in Bengaluru. The company’s pitch is aimed at borrowers building homes in rural and semi-urban markets — people who often have fragmented income proof, thin documentation, or simply no easy path into formal mortgage underwriting. That’s why Nivasa talks about itself less like a pure lender and more like infrastructure that helps lenders enter a hard-to-serve segment with less friction.

Founder market fit

Shetty is the more obvious operator for the distribution and lending side. Before Nivasa, he was vice president of strategy and M&A at Navi Technologies and CEO of Navi Finserv, where he worked on digital personal and housing loans. Earlier, he founded and ran Chaitanya India Micro Finance, later acquired by Navi, and he studied at IIM Ahmedabad after completing an engineering degree from Bangalore University.

Saraf brings the credit brain. He leads credit policy and Nivasa’s lender allocation engine. Before this, he built and ran credit-risk functions at SmartCoin and ZestMoney, where he worked on AI- and ML-based risk engines and automated decisioning systems. He has 18 years of experience across fintechs, multinational banks, and credit bureaus, and studied mathematics at IIT Kanpur. A strong fit.

Early traction and fundraising

Nivasa is already live, not just testing slides. It has partnered with more than 10 lending institutions across banks, small finance banks, housing finance companies, and NBFCs, and the source article says it has disbursed more than ₹20 crore while piloting the model in Mysore and Mandya. Its own properties also show operating branches in Mysuru, Mandya, and Ramanagara. That suggests the company is building an actual field footprint instead of staying purely digital.

The new round is a seed round of ₹25 crore. Prime Venture Partners led it, with Blume Ventures, Whiteboard Capital, and angel investors also participating. Nivasa will use the money over the next 12 months to expand geographically, strengthen its distribution network, deepen partnerships with banks, NBFCs, and HFCs, and invest harder in field execution. It’s also exploring an NBFC licence, which would move it closer to the lending stack instead of remaining only a distribution intermediary.

Competition and market positioning

Here’s where Nivasa gets interesting. It doesn’t sit neatly beside large affordable housing finance companies such as Aadhar Housing Finance, Aavas Financiers, Home First, Aptus Value Housing Finance, or India Shelter, because those firms actually lend off their own books. It’s also different from Easy Home Finance, which is a mortgage-tech lender with its own lending operations and broader pan-India expansion plan.

Nivasa’s closer comparison is the old patchwork it’s trying to replace: local DSAs and branch-level sourcing. Manual screening, too. And lender shopping done through personal networks. Its edge, if it works, is speed plus control — a digital front end, a credit-aware matching engine, and a field network that can still show up at a customer’s doorstep. Investors are backing that distribution-first approach in secured lending, especially in markets where formal mortgage supply exists but origination and fulfilment are still broken. That’s a smart bet. But it’s still a bet.

Why does Nivasa Finance’s ₹25 Cr seed round matter?

Seed rounds in lending-adjacent businesses usually go one of two ways. Either the money disappears into customer acquisition with no moat, or it funds the messy operational layer others don’t want to build. Nivasa looks more like the second case.

The company isn’t using this round to chase vanity scale. It’s putting capital into geography and lender relationships. Field execution is in there too. Those are the exact pieces that determine whether a home-loan fulfilment model can work outside top metros. For borrowers, that means more local access points and faster case movement. For lenders, it means better reach into rural and semi-urban demand without building every last distribution pipe themselves.

And the NBFC angle matters too. If Nivasa eventually secures that licence, it gets the option to move from being a facilitator into owning more of the lending economics. That doesn’t happen overnight, and it adds regulatory weight. But it does show ambition beyond being “just another DSA” with a nicer app.

How big is India’s affordable housing finance market?

The timing isn’t random. India’s individual housing finance market is currently valued at about ₹33 trillion and is projected to reach ₹77 trillion to ₹81 trillion by FY30, implying a 15% to 16% CAGR over FY25 to FY30. CareEdge also noted that residential property sales were up 74% from CY19 to 4.6 lakh units in CY24. That helps explain why investors keep coming back to housing credit.

Inside that, affordable housing finance is still one of the more active pockets. Sector estimates point to affordable housing finance companies growing assets under management by 20% to 21% in FY26 and FY27, while another sector report pegs the AHFC market at ₹14.1 trillion by FY27. That’s big enough to matter. Still fragmented enough for newer models to carve out room.

There’s a real structural shift underneath this. More lenders are willing to underwrite borrowers with informal or semi-formal income if the sourcing, documentation, and early risk filters are tighter. Government support for affordable housing and the push from PMAY 2.0 also help demand hold up in lower-income segments. That doesn’t remove credit risk — some analysts have already flagged mild stress in smaller-ticket loans — but it does explain why investors are still willing to back distribution, underwriting, and fulfilment plays in this category.

Will Nivasa Finance become more than a home loan distributor?

Nivasa Finance has a clean story: fix the part of housing credit that breaks before the loan ever gets booked. That’s a real problem, and the company already has enough founder depth and early traction to make the story credible.

What to watch next is execution, not branding. Can it expand beyond its Karnataka pilot markets without losing quality? Can it keep lenders happy while building a much larger field network? And if the NBFC plan becomes real, can Nivasa Finance make the jump from being a strong fulfilment layer to becoming a bigger secured-lending business in its own right?

Read how Dessn raised $6M led by Connect Ventures to help product teams prototype directly inside real codebases instead of relying on traditional design-to-engineering handoffs.

FAQ

– What funding did Nivasa Finance raise?  

 Nivasa Finance raised ₹25 crore in a seed round announced on May 13, 2026. Prime Venture Partners led the round, and Blume Ventures, Whiteboard Capital, and angel investors also joined in.

– How does Nivasa Finance work for a home-loan customer?  

 Nivasa Finance acts as an intermediary between borrowers and formal lenders rather than lending directly from its own balance sheet. A customer can begin through WhatsApp or the company’s portal, go through digital onboarding and assessment, and then get matched to a suitable bank, NBFC, small finance bank, or housing finance company.

– Who are the founders of Nivasa Finance?  

 Nivasa Finance was founded in 2025 by Samit Shetty and Hitesh Saraf. Shetty previously held senior roles at Navi and built Chaitanya India Micro Finance, while Saraf spent years building credit-risk systems at SmartCoin and ZestMoney.

– Is Nivasa Finance an NBFC or a housing finance company?  

 Right now, no — Nivasa Finance operates as an intermediary and direct selling agent for RBI-authorized lenders. The company is exploring an NBFC licence, which could let it play a deeper role in the secured lending stack later on.

Dessn is an AI-native product design platform that lets teams prototype directly inside an existing codebase instead of mocking things up somewhere else first. The Dessn design tool has now raised $6 million, led by Connect Ventures, as more startups look for faster ways to close the gap between design and shipped software. Founded in 2024 by Gabriella Hachem and Nim Cheema, the company is betting that the old handoff between Figma files and engineering tickets is starting to look slow — and a little outdated.

That’s the pitch.

And it’s a sharp one. Dessn isn’t trying to replace brainstorming tools or one-shot prompt apps. It wants to own the messy middle — where real product teams already have code, real constraints, and too many rounds of “can engineering make this actually work?”

What is the Dessn design tool and how does it work?

The cleanest way to explain Dessn is this: a company gives the platform read access to its repository, and Dessn spins up a cloud design environment around that codebase so designers and product managers can prototype in the real product context without opening an IDE. Every prototype runs as a live branch of the codebase. That means the team is working against actual components and system behavior instead of an approximation. Setup is basically one click, with environments targeted to be ready within 24 hours.

That matters because the product isn’t just a text box that spits out mockups. Dessn supports React web apps, and teams don’t need to clean up their stack first — it works across existing CSS choices and component libraries. Each project runs in an isolated microVM. The repo stays read-only, and Dessn never writes or pushes code back automatically. That’s a direct answer to the trust problem that kills a lot of “just connect your codebase” products.

There are also some more interesting workflow touches. Nim Cheema has described Dessn as combining visual component rendering with AI-driven code help. Earlier posts from the company point to features like visual search — navigating a codebase from a screenshot of the live app, a Figma mockup, or even a rough sketch. That gives a clearer picture of what the product is trying to be: less “AI design toy,” more shared interface between design intent and production software.

Before Dessn, a designer might make static mocks, wait for engineering bandwidth, then discover the real app behaves differently. After Dessn, the prototype lives where the product already lives. The company offers a free tier that lets teams compile 1 repository and use 5 prompts a week. Paid plans start at $39 per user per month. Higher tiers unlock more prompts, public links, and the option to opt out of AI training.

Who founded Dessn and why are they building it?

The founding story

Dessn was founded in 2024 by Gabriella Hachem and Nim Cheema, who had already worked together at 2 previous startups. Their core idea was simple and pretty provocative: if code keeps getting cheaper to generate, design becomes more valuable as the differentiator. Cheema summed up that thesis bluntly when he said they started from the belief that “code is going to get commoditized.”

That idea helps explain why Dessn is so focused on teams with existing products. This isn’t a blank-canvas ideation app like Lovable or Vercel’s v0. It’s built for companies that already have software in market and want to iterate on the real thing faster.

Why the founders fit this problem

Hachem comes from product, UX, and marketing roles across SaaS, enterprise, e-commerce, and B2B/B2C software. Before Dessn, she worked as a product manager at Planned and Automat.ai, and also held a conversational AI UX design role at Automat.ai. She studied marketing and entrepreneurship at McGill.

Cheema brings the more technical half. He started as a software engineer, spent time in data visualization, and moved toward product engineering with a user-experience focus. Before founding Dessn, he was involved in AI work and later held the Head of AI title at Planned. The pairing is coherent: one founder from product and UX, the other from engineering and AI, both circling the same handoff problem from different sides.

Early traction, pricing, and team size

Dessn is already live with real customers. Named users include teams at Color, Wispr, and Mercury. On Dessn’s own customer example for Color, 104 team members have used the platform over 10 months, creating 421 prototypes, with the largest prototype involving 570 prompts. In separate funding materials, Dessn said some users spend more than 5 hours a day in the product. The company itself is still tiny, with 4 people, and plans to add only a few more.

The $6M round and what investors are buying

Connect Ventures led the startup’s new $6 million round, with participation from Betaworks and N49P. Other reporting says the amount combines a seed round with a small, previously unannounced pre-seed. The money is expected to go toward hires and growth rather than a giant headcount sprint.

Investors aren’t just buying revenue hopes here. They’re buying a different product theory. Betaworks partner Jordan Crook argued Dessn is the kind of tool Figma would build if it were starting from scratch today — one centered on production fidelity instead of translating designs into code after the fact. It gets at why this round stands out. Dessn is trying to shift the design surface itself.

How the competition looks from here

The obvious comparison set is the current wave of AI design tools — Visual Electric, Weavy, Flora, and Krea. But those products mostly focus on concept generation, media creation, or AI-assisted canvas workflows. Visual Electric was acquired by Perplexity in October 2025 after raising $2.5 million. Weavy was acquired by Figma and turned into Figma Weave, which now emphasizes browser-based image, video, and motion workflows. Flora raised $42 million in January 2026 to expand its AI-native creative workflow product. Krea has grown into a heavily funded image-and-video creation company with $83 million in backing.

Dessn’s angle is narrower and, frankly, more opinionated. It doesn’t want to be the place where you dream up anything from nothing. It wants to be the place where a team with an existing app changes real flows and real components. Real interfaces too. Hachem has also made a point of saying Dessn doesn’t create switching costs — teams can use it alongside Figma — but it deliberately doesn’t want a Figma integration because that would pull work away from production instead of into it.

Why this Dessn design tool funding round matters

This round matters because Dessn has already chosen a hard path.

A lot of AI product tools stay vague on infrastructure. Dessn didn’t. It built around the ugly part first — running different customer codebases in the cloud without needing a developer to do setup. If that works reliably, the company can become more than another prompt interface. It becomes workflow plumbing for product teams.

The money should also help Dessn push into the next layer of collaboration. Right now the product has no integrations. The roadmap points toward tools like Slack and meeting notetakers such as Granola, where a discussion could turn straight into a prototype. That sounds ambitious, and maybe a little dangerous — lots of AI tools get bloated fast — but it fits the founders’ worldview that product decisions should turn into working artifacts much faster than they do now.

There’s also a more philosophical piece here. Hachem said she and Cheema are “token maximalists,” meaning they’d rather spend more compute to reach the right result than preserve a static UI just to look familiar. So if you’re expecting a classic toolbar-heavy design app, Dessn probably won’t go there. This funding gives the company room to keep that view intact instead of sanding it down too early.

What market shift is creating demand for design-in-production tools?

The market backdrop is favorable for a product like this. Grand View Research estimates the global CAD software segment was worth about $11.2 billion in 2024 and projects it to reach roughly $33.0 billion by 2030, with a 20.5% CAGR. That’s not a direct proxy for AI product design tools, but it does show how quickly software for design, modeling, and digital product work is expanding.

The workflow trend lines are moving in Dessn’s direction. An ASME report on CAD in 2030 said 80% of end users in the U.S. prefer cloud-hosted and SaaS applications for communication and organization. Separate 2026 research on generative AI in software engineering found 79% of developers use GenAI daily, and more than 70% reported at least halving time on tasks like boilerplate and documentation. When teams are already comfortable with cloud tools and AI-assisted software work, the leap from “design file first” to “production context first” starts to look a lot less weird.

That doesn’t mean Dessn automatically wins.

It does mean the company is launching into a moment when code is becoming easier to generate, browser-based creative tools are normal, and more product teams are willing to let non-engineers get closer to the software itself.

The takeaway on the Dessn design tool

The interesting thing about Dessn isn’t just the $6 million. It’s that the company is making a pretty direct attack on the old idea that design has to happen somewhere separate from the product.

If the Dessn design tool can keep setup light, expand beyond early customers, and add collaboration layers without losing its production-first discipline, it could end up mattering a lot more than most AI design startups. What to watch next is simple: integrations, broader technical support, and whether teams decide they really want design to happen inside the codebase instead of next to it.

Read how Exaforce raised a $125M Series B to build an AI-agent-powered SOC platform that helps security teams automate threat detection, triage, investigation, and response.

FAQ

– What funding did Dessn raise? 

 Dessn raised $6 million in funding in May 2026. Connect Ventures led the round, and Betaworks plus N49P also participated; separate reporting says the total included a seed round and a smaller pre-seed that hadn’t been announced before.

– How does the Dessn design tool work? 

 Dessn connects to a company’s repo with read access and creates a cloud environment where teams can prototype directly against the real codebase. It’s built for React web apps today, uses isolated microVMs, and keeps the repo read-only so teams can explore changes without automatically writing back to production code.

– Who are the founders of Dessn? 

 Dessn was founded in 2024 by Gabriella Hachem and Nim Cheema. Hachem worked in product and UX roles at Planned and Automat.ai, while Cheema came up through software engineering, data visualization, product engineering, and AI work before helping start the company.

– Is Dessn a Figma competitor or an AI developer tool? 

 It’s a bit of both, but not in the usual way. Dessn sits closer to an AI product-design workflow tool for teams with existing software, while tools like Figma Weave, Flora, Visual Electric, and Krea are more centered on generative creation, editing, or ideation rather than working directly inside a production codebase.

Exaforce builds software that uses AI agents to automate security operations work inside the modern SOC. The Exaforce SOC platform just pulled in a $125 million Series B at a $725 million valuation, a big jump that shows investors think security teams will spend serious money on tools that cut through alert noise fast. The problem is simple: false positives and repetitive triage bury analysts, which is why Umesh Padval of Seligman Ventures compared the job to hunting for a needle in a haystack. Founded in 2023, Exaforce is led by CEO Ankur Singla and co-founders Ariful Huq, Jakub Pavlik, and Devesh Mittal.

The timing isn’t subtle. A year after landing a $75 million Series A, Exaforce is back with a much larger round, bringing total funding to $200 million. Singla’s pitch is blunt: use AI to catch and stop threats as they happen. “It’s a very simple mandate, but it’s very complex to execute,” he told TechCrunch.

What is Exaforce SOC platform and how does it work?

The Exaforce SOC platform is an agentic security operations system that pulls together security data, analyzes it, and moves teams from alert to decision much faster than the usual SIEM-plus-manual-investigation workflow. Exaforce splits that work across four task-specific agents: Exabot Detect, Exabot Triage, Exabot Investigate, and Exabot Respond. Those bots run on a unified, real-time view of the environment. Customers can use the software directly or buy it as a managed detection and response service.

Under the hood, Exaforce isn’t pitching a single-LLM chatbot for analysts. It uses what it calls a multi-model AI engine that combines semantic models with behavioral and statistical models. It also uses LLM-based reasoning. The point is to work with the messy stuff security teams already deal with — logs, cloud telemetry, identity data, third-party alerts, source code, files, folders, and AI tool usage data — without asking humans to stitch the whole story together by hand.

That changes the workflow in practical ways. The platform can automatically triage alerts and enrich them with context. It also surfaces attack paths and helps threat hunters run hypothesis-driven investigations in plain English. Exaforce’s newer “vibe hunting” feature builds on that idea: analysts can ask whether new attacks appear to be coming from Iran and start investigations from a hunch instead of relying on rigid query language.

It also reaches into response. Exaforce says the system can handle chores like resetting MFA and killing user sessions. It can also disable devices, confirm actions with users or managers, and draw on prior ticket history during new investigations. The company puts the reduction in manual, time-consuming work at as much as 90%. That’s an aggressive number.

Who founded Exaforce and why are they credible?

How Exaforce got started

Exaforce was founded in 2023 by Ankur Singla, Ariful Huq, Jakub Pavlik, and Devesh Mittal. The company’s pitch from day 1 was that security teams needed more than another AI assistant bolted onto legacy tools — they needed one platform that could cover detection, triage, investigation, and response. By April 2025, Exaforce was already framing that vision as a 10x productivity push for SOC teams, and by Q4 2025 it had formally brought the product to market after 2 years working with design partners.

Why the founders fit this category

Singla isn’t new to security infrastructure. He previously held roles at F5, Juniper, and Cisco, and the founding bench has operated at companies including Google, F5, and Palo Alto Networks. Pavlik’s background is especially relevant for the operational side: before Exaforce, he led SRE and security operations work tied to F5XC and Volterra. Earlier, he helped build private cloud company tcp cloud.

That matters because Exaforce isn’t selling a generic AI layer. It’s trying to automate the ugliest, highest-context parts of security operations. That’s hard. Huq’s background spans product, engineering, and technical operations, which helps explain why Exaforce talks as much about workflow and usability as it does about models.

Past ventures, traction, and fundraising

There’s some repeat-founder signal here too. Singla and Pavlik previously worked together at Volterra, which F5 acquired, and Pavlik was also involved in tcp cloud, which sold to Mirantis. That doesn’t guarantee a win this time. But it does mean investors aren’t betting on first-time founders learning enterprise security sales, infrastructure, and platform design from scratch.

On traction, Exaforce is still early but no longer in science-project mode. It launched commercially in Q4 2025 after testing with enterprise design partners, and it has added 20 customers so far, including Replit and Guardant Health. Singla told TechCrunch he expects that figure to reach 40 to 50 customers by the end of 2026.

The new money is the headline: $125 million in Series B funding at a $725 million valuation, with participation from HarbourVest, Peak XV, Mayfield, Khosla Ventures, and Seligman Ventures. Exaforce raised $75 million in Series A in April 2025, so total funding now sits at $200 million. A round of this size is meant to fund heavy product development and the expensive enterprise go-to-market motion required to sell an end-to-end AI SOC platform.

Competition is getting crowded fast. TechCrunch named 7AI, Dropzone AI, and Prophet Security as direct startup rivals, while Palo Alto Networks and CrowdStrike loom as giant incumbents. Exaforce is trying to sell breadth: not just AI triage, but a full workflow spanning detection through response. It also pitches a data layer that correlates cloud, SaaS, identity, endpoint, and email signals. Dropzone AI raised a $37 million Series B in July 2025 and says it serves more than 100 enterprises; Prophet Security raised a $30 million Series A that same month; and 7AI raised $130 million in December 2025 at a $700 million valuation. Investors are piling into AI SOC tooling. Exaforce won’t have much room for execution mistakes.

Why does Exaforce funding round matter?

A round this size matters because Exaforce isn’t building a lightweight add-on. It’s trying to replace a messy stack of alerting and investigation workflows. It also wants to cover hunting and response with one AI-native system. That takes a lot of engineering, a lot of integrations, and a lot of customer hand-holding. The $125 million raise looks less like vanity financing and more like acknowledgment that enterprise security buyers want proof, precision, and support before they trust automation in high-stakes environments.

For customers, the signal is clear too. Singla said recent attacks have “supercharged” Exaforce’s path into accounts, and that buyers are no longer asking why they need automation so much as how to operationalize it. That’s a shift from curiosity to deployment.

But money alone doesn’t settle the question. Buyers will still care about whether the platform is accurate, explainable, and safe when it starts taking response actions. Exaforce’s human-in-the-loop language and multi-model architecture are meant to answer that concern. Now it has to prove those design choices hold up outside early adopters.

How big is the AI SOC market getting?

The macro setup is favorable. Gartner said enterprise spending on cybersecurity software and network security would grow 14% in 2025 to $118.5 billion. That doesn’t map one-to-one to the AI SOC category, but it shows how much budget is already flowing into tools that help enterprises defend increasingly complex environments.

The labor problem hasn’t gone away. ISC2’s 2024 Cybersecurity Workforce Study was based on 15,852 respondents, and its findings showed a profession under strain: 58% said staffing shortages pose significant risk to their organizations, while 64% said skills gaps can hurt security even more than pure headcount shortages. That’s why startups like Exaforce, Dropzone, Prophet, and 7AI keep getting funded. The pitch isn’t just “AI is cool.” It’s that the old model of throwing more analysts at more alerts has stopped penciling out.

Final take on Exaforce SOC platform

Exaforce has gone from stealthy infrastructure bet to one of the better-funded companies in agentic security operations in just 3 years. The Exaforce SOC platform is ambitious — maybe uncomfortably ambitious — because it’s trying to automate the full loop, not just the easy first step. That’s why investors are interested.

What to watch next is simple: customer growth. Exaforce says it has 20 customers today and wants 40 to 50 by the end of 2026. If it hits that while keeping response quality high, this won’t look like just another big AI security round. It’ll look like early proof that the SOC is being rebuilt around agents.

Read how Dil Foods raised ₹72 Cr in Series B funding led by Bikaji Foods Family Office to turn underused local kitchens into scalable delivery-first food brands across India.

FAQ

– What is the latest Exaforce funding round? 

 Exaforce raised a $125 million Series B on May 12, 2026, at a $725 million valuation. The round included HarbourVest, Peak XV, Mayfield, Khosla Ventures, and Seligman Ventures, and it came roughly a year after the company’s $75 million Series A, bringing total funding to $200 million.

– How does Exaforce SOC platform work? 

 It works by combining a unified security data layer with 4 AI agents — Detect, Triage, Investigate, and Respond — that handle different parts of the SOC workflow. Exaforce also uses a multi-model architecture, not just a single LLM, so it can reason across logs, identity data, cloud telemetry, third-party alerts, and response actions with more context than a chatbot-style copilot.

– Who founded Exaforce? 

 Exaforce was founded in 2023 by Ankur Singla, Ariful Huq, Jakub Pavlik, and Devesh Mittal. Singla brings experience from F5, Juniper, and Cisco, while Pavlik previously worked on Volterra and tcp cloud, giving the company a founding team with real infrastructure, cloud, and security operations depth.

– Is Exaforce part of AI cybersecurity or broader enterprise software? 

 It’s both, but the cleaner label is AI cybersecurity — more specifically, agentic security operations software. Exaforce sits in the emerging AI SOC category, where vendors try to automate alert triage, investigation, threat hunting, and response for security teams that can’t keep scaling headcount at the same pace as alert volume and cybersecurity spending.