Grishin Robotics

Nexus (YC F25) has raised $4.3M in seed funding led by General Catalyst, with participation from Y Combinator, Transpose Platform, Twenty Two Ventures, Phosphor Capital, and angel investors including Gokul Rajaram, Raphael Schaad, and Jake Mintz. The company is building an enterprise AI agent platform that lets non-technical teams deploy autonomous agents across core systems, which makes this a bet on operationalizing agentic AI inside real enterprises rather than leaving it in pilot mode. Most enterprise AI products still stall between demo and deployment because the integration work, governance requirements, and change-management overhead sit outside the model itself. Nexus is trying to close that gap with a no-code layer across 4,000+ systems plus hands-on rollout support, and the early customer signals are unusually concrete: Orange says a customer onboarding agent went live in weeks, lifted conversion by 50%, and generated more than $6M in annual lifetime value, while Lambda is already using the platform across sales and marketing workflows. The more enterprise buyers shift from assistants that suggest work to agents that complete work inside governed systems, the more value may accrue to platforms that own deployment speed, integration depth, and organizational adoption together. Quick facts👇 ● founders: Assem Chammah; Shady Al Shoha ● total capital raised: $4.3M ● HQ: Brussels, Belgium ● Investors: General Catalyst; Y Combinator; Transpose Platform; Twenty Two Ventures; Phosphor Capital; Gokul Rajaram; Raphael Schaad; Jake Mintz

Alcatraz just raised $50M in Series B funding led by BlackPeak Capital, Cogito Capital, and Taiwania Capital, bringing total capital raised to more than $100M. The company is building a privacy-first facial authentication layer for physical access control, which makes this less like another biometrics round and more like a bet that AI infrastructure now needs identity-native security at the door. Most physical security systems still depend on badges, PINs, or legacy facial recognition stacks that create both friction and privacy risk. Alcatraz is taking a different route: on-device facial authentication that verifies identity in real time without storing biometric images or personal data. That product positioning is landing in exactly the environments where the cost of a physical breach keeps rising - AI data centers, airports, energy sites, and Fortune 500 campuses. Reported 2025 growth of 300% in data center adoption, 200% in enterprise customers, and 5x expansion across Fortune 500 deployments suggests this is not just a compliance story. It is infrastructure timing. As hyperscalers keep pouring capital into AI facilities, the companies securing the physical perimeter may become part of the AI stack too. If compute is the engine, access control is starting to look like a system-level dependency. Quick facts👇 ● founders: Vince Gaydarzhiev ● total capital raised: more than $100M ● HQ: Cupertino, California ● Investors: BlackPeak Capital; Cogito Capital Partners; Taiwania Capital; Almaz Capital; EBRD; Ray Stata

Soma Energy has raised $7M in seed and pre-seed funding led by Category Ventures to help data centers get to power faster. The company is applying AI to a problem that is becoming central to the AI stack itself: coordinating generation, storage, and large loads in real time so compute capacity is not stranded by grid bottlenecks. Most infrastructure conversations around AI still focus on chips, models, and new power buildout. Soma is going after the shorter-term constraint - time to power. Its platform sits across supply and demand, helping independent power producers trade existing assets more intelligently while giving data centers a control layer that can turn on-site generation, batteries, and load into flexible grid assets. With about 2 GW already under optimization for power-producing clients and five data center customers in flight, this looks less like energy software for its own sake and more like enabling infrastructure for AI deployment. As AI data center demand keeps climbing, companies that can unlock usable megawatts faster may matter as much as those building new capacity. Quick facts👇 ● founders: Athanasios Caramanolis; Mario Souto; Henrique Helfer Hoeltgebaum Hoeltgebaum ● total capital raised: $7M ● HQ: Vancouver, British Columbia ● Investors: Category Ventures; Haystack; RRE Ventures; TO VC; Uncork Capital; Panache Ventures; Walter Kortschak ● Partners: H5 Data Centers (2026)

daydream just raised $15M in Series A funding led by WndrCo, with First Round Capital and Basis Set Ventures participating. The company is building an AI-native SEO agency that combines agents with human operators, which makes this less like another marketing services round and more like a bet on software-driven execution inside a large, messy channel. Most SEO platforms still stop at dashboards, recommendations, and visibility reports. daydream is trying to operationalize the work itself across keyword strategy, technical fixes, on-page execution, programmatic workflows, and AI visibility, while keeping a Growth Lead in the loop for judgment and accountability. That hybrid model fits a moment when generative AI is changing how traffic gets discovered, but most companies still do not have an execution system built for that shift. Y Combinator recently ranked AI-native agencies among the startup categories it most wants to fund, and this round looks like a direct expression of that thesis. If this category keeps working, some agency businesses may start to look a lot more like product companies with services wrapped around them. Quick facts👇 ● founders: Thenuka Karunaratne; Shravan Rajinikanth ● total capital raised: $21M ● HQ: San Francisco, California ● Investors: WndrCo; First Round Capital; Basis Set

Galaxea just added about $290M in Series B+ funding at a reported $29B valuation, only weeks after a separate roughly $140M Series B round. That pace says something about where embodied AI capital is going in China: toward teams trying to close the gap between model demos and robots that can actually execute in messy physical environments. Galaxea's bet is not simulation-first scale. The company says it is building Vision-Language-Action and World Action models on real-world embodied data, a harder and more expensive path than synthetic-heavy training but one that could matter if reliability in open environments is the real bottleneck. That also helps explain why founders coming out of Waymo, Tsinghua, and Momenta are getting so much attention here - they are building around data loops, hardware, and deployment, not just model benchmarks. The more embodied AI shifts from research theater to repeatable field performance, the more capital will concentrate around teams that can collect proprietary real-world data faster than rivals. Can a real-data training strategy beat simulation-heavy rivals as embodied AI moves into commercial deployment? Quick facts👇 ● founders: Jiyang Gao; Hang Zhao; Tianwei Li; Huazhe Xu ● total capital raised: ≈$430M disclosed ● HQ: Beijing, China ● Investors: industrial investors; long-term funds; state-backed capital; Baidu Venture; GSR Ventures; IDG Capital; Cathay Innovation

Numos just raised $4.25M in seed funding led by General Catalyst, with Operator Collective 🔆 participating. The company is building AI agents for finance teams that sit on top of existing systems and show their work, which matters in a function where trust, auditability, and source traceability decide whether AI gets used at all. Most enterprise finance software still moves data around and leaves humans to stitch together context across ERP systems, billing tools, warehouses, and spreadsheets. Numos is going after that translation layer with explainable agents for variance analysis, reconciliations, quote-to-cash, and close workflows. Early customer names like Udemy, Dandy, Tarro, and Verdigris suggest there is real appetite for AI that can speed finance operations without turning the office of the CFO into a black box. The more AI moves into controllership and FP&A, the more the winners may look less like copilots and more like auditable systems of action built around financial truth. Quick facts👇 ● founders: Parijat Sarkar; Mitul Tiwari ● total capital raised: $4.25M ● HQ: San Mateo, California ● Investors: General Catalyst; Operator Collective 🔆

Moonbounce just launched with $12M led by Amplify Partners and StepStone Group to build what is effectively a runtime control plane for AI behavior. Most AI safety tooling still lives upstream in model training or downstream in manual moderation. Moonbounce sits in the decision loop itself, turning policy into real-time enforcement across chatbots, image generators, and user-generated platforms. That matters because the failure mode in generative AI is no longer just bad output quality - it is liability, compliance, and trust at production scale. The early usage signal is strong for a company at launch: more than 1 trillion tokens processed, about 50 million content evaluations per day, and deployments across products like Civitai and Dippy. Brett Levenson and Ash Bhardwaj are also selling from first-hand experience inside Meta and Apple, which gives Moonbounce more credibility than a generic AI guardrails pitch. As AI products move from demo to daily habit, control infrastructure may become as important as the models themselves. Quick facts👇 ● founders: Brett Levenson; Ash Bhardwaj ● total capital raised: $12M ● HQ: Oakland, California ● Investors: Amplify Partners; StepStone Group; PrimeSet; Josh Leslie

Israel Duanis and Niv Goldenberg first connected in Israel's elite Talpiot program more than 20 years ago. Their company, Linx Security, just closed a $50M Series B led by Insight Partners, bringing total funding to $83M. Linx governs every identity operating inside an enterprise: human users, machine accounts, and now AI agents. Its recently launched Autopilot agent monitors identity activity in real time and takes automated remediation action. The company has already secured multimillion-dollar contracts with banks, healthcare organizations, and Fortune 500 companies managing millions of identities globally. Non-human identities already outnumber human users in most organizations. AI agents are adding a new category that inherits permissions originally designed for people, expanding the attack surface faster than security teams can track. Revenue grew 10x over the past year, a signal that enterprises are treating identity governance as a real-time security problem rather than a periodic compliance checkbox. Identity governance spent two decades as a back-office compliance function. AI agents may be the forcing function that turns it into a front-line security discipline. Quick facts👇 ● founders: Israel Duanis; Niv Goldenberg ● total capital raised: $83M ● HQ: New York, New York ● Investors: Insight Partners; Cyberstarts; Index Ventures

Anvil Robotics wants to be the AWS for physical AI. Eight months after launching, the San Francisco startup has shipped over 100 robots to 50+ customers including Nvidia's GEAR lab and Path Robotics, and just closed a $5.5M seed led by Matter Venture Partners. The pitch is straightforward: physical AI teams spend six-plus months gluing together arms, cameras, and open-source libraries just to get a prototype running. Anvil ships composable, open-source robot devkits that work out of the box in 1-2 days, starting at $1,900. Think modular hardware platform with integrated software, controls, and data tools. Matter incubated Anvil as a "robotics foundry" - founding partner Haomiao Huang compared the model to what TSMC did for chips and what AWS did for SaaS. The startup controls its own manufacturing in Taiwan and open-sources all robot designs, which means customers can customize hardware without vendor lock-in. Robotics startup funding hit a record $14B in 2025. Most of that capital is flowing to companies building proprietary, single-purpose systems. Anvil is making the opposite bet - that an open, composable platform wins by giving teams building blocks instead of finished products. Quick facts👇 ● founders: Mike X.; Vijay Pradeep ● total capital raised: $6.5M ● HQ: San Francisco, California ● Investors: Matter Venture Partners; Humba Ventures; DNX Ventures - US; Vivek Sodera; Spacecadet; Position Ventures

The workforce management software market is on track for $9.76 billion this year, and most of it still runs on tools designed before the iPhone. Sona just raised $45M in Series B funding, led by N47, to fix that. Sona's platform pulls scheduling, HR, payroll, compliance, and analytics into one stack, then layers AI demand forecasting on top, using real-time signals like bookings, revenue, and weather to predict staffing needs. Early adopters report generative AI copilots cutting schedule creation time by up to 70%. The competitive landscape tells the story. UKG, Ceridian's Dayforce, Workday, Deputy, and ADP collectively hold 55% of the global market. All are legacy platforms retrofitting AI onto architectures built for a different era. Sona is building AI-native from the ground up, with customers like Popeyes and Tao Group already on board. Its newest product, Forge, lets businesses build custom internal software on the platform, moving Sona from a point solution toward operational infrastructure. Workforce management is one of the last enterprise categories where AI-native has not yet won. If Sona's US expansion lands, the same displacement pattern playing out across SaaS hits another vertical. Quick facts👇 ● founders: Steffen Wulff Petersen; Ben Dixon; Oli Johnson ● total capital raised: >$100M ● HQ: London, England ● Investors: N47; Felicis; Northzone; Gradient; Italian Founders Fund