Agentic AI's Next Frontier Is Not Your Inbox. It's a CNC Machine.

Every AI funding story this year has involved some variation of the same plot: smart people build a large language model wrapper for knowledge workers, raise tens of millions of dollars, and promise to eliminate someone's email inbox. Limitless Labs, a Tel Aviv-based startup formerly known as LimitlessCNC, has a different pitch. Its agent does not write your Slack messages. It programs CNC machines.

The Round and

What It Tells You Limitless Labs closed a $20 million Series A co-led by Dell Technologies Capital and Square Peg, with additional participation from Grove Ventures, Meron Capital, and Kinetica, according to FinSMEs. The raise brings the company's total funding to $27.3 million. That is not a typo; the pre-Series A rounds were small, which is either a sign of capital efficiency or a sign that most investors looked at the words "CNC programming" and quietly closed the tab. Dell Technologies Capital and Square Peg looked at those same words and wrote the lead check, which is the part worth paying attention to. Founded in 2024 by David Priev, Assaf Peleg, and Shahaf Finder, the company is two years old and already in production with Blue Origin, the Cadillac Formula 1 Team, Sandvik, and Iscar, per ManufacturingTomorrow. Aerospace, defense, motorsports, and industrial machinery are the named verticals. These are not pilot customers who agreed to a free trial in exchange for a case study. These are organizations where a machining error is not a bug ticket; it is a scrapped titanium part worth more than your laptop.

What the Agent Actually Does

The technical premise is worth understanding carefully, because this is where "agentic AI" stops being a buzzword and starts being a specific engineering decision. According to Robotics and Automation News, the platform works inside CAM software that engineers already use, specifically Siemens NX CAM, Mastercam, and PTC Creo, as confirmed by FinSMEs. The agent does not replace those tools. It sits inside them, which is roughly the same architectural move that Copilot made by embedding inside Visual Studio Code rather than asking developers to switch editors. The company describes its goal as helping manufacturers capture, standardize, and scale the expertise of their most experienced programmers, per HPC Wire. That framing is important. CNC programming is a craft skill. Experienced machinists encode decades of intuition into toolpaths, feed rates, and fixture choices. When those people retire (and they are retiring at a brisk pace across Western manufacturing), that knowledge largely walks out the door with them. The agent is essentially a knowledge-capture and replay system with autonomy layered on top. The SaaS News reports the platform can reduce CNC programming time by up to 50 percent.

Where the Money Goes

The use-of-funds disclosure is unusually specific, which makes it more useful than the generic "accelerate growth and expand the team" boilerplate that fills most funding announcements. According to FinSMEs, Limitless Labs plans to build out a dedicated U.S. commercial organization, advance its Physical AI Foundation Model toward closed-loop CNC automation, and grow its CAM Agent. The phrase "closed-loop" is doing real work in that sentence. It means the system eventually observes outcomes from the physical machine, feeds that data back, and adjusts future programs accordingly. That is a meaningfully harder problem than generating a toolpath from a CAD file, and it is the direction the company is signaling. The platform is currently deployed across cloud, private VPC, and AWS GovCloud environments with full ITAR compliance support, per FinSMEs. The GovCloud and ITAR details are not incidental. Defense manufacturing is one of the named verticals, and ITAR compliance is a prerequisite for working with U.S. defense contractors. That infrastructure choice is the company telling defense primes, quietly, that they can take the call.

Why This Matters Beyond

the Factory The broader point for people building with AI is what this bet implies about where agentic systems actually earn their keep. The knowledge-work case for agents relies on a forgiving environment: if the agent drafts a bad email, you edit it. The manufacturing case is less forgiving by several orders of magnitude. A bad toolpath crashes a spindle or gouges a part. The fact that Limitless Labs is already in production at Blue Origin and Cadillac F1 (per ManufacturingTomorrow) suggests the agent has cleared a tolerance bar that most AI deployments never have to face. For builders and researchers watching the agentic AI space, the Limitless Labs raise is a useful reference point. Physical environments with high consequences, specialized domain knowledge, and existing software toolchains are exactly the settings where a well-designed agent adds durable value rather than novelty value. The inbox will still be there. The CNC machine is more interesting.