The most expensive person in a factory is often the one who knows how to talk to the robots. They are the specialists, often contractors, who write the thousands of lines of code that tell an industrial arm how to pick up a part, move it, and place it with sub-millimeter precision. Their time is measured in billable hours, and their scarcity is a primary bottleneck in deploying automation, especially for tasks that involve any variability. Trener Robotics, a startup out of San Jose and Trondheim, thinks it can replace that person with a sentence.
Their platform, Acteris, is pitched as a foundational intelligence layer for industrial robots. In practice, it is a library of pre-trained AI skills,like seeing, grasping, and placing,that a factory operator can summon and configure using natural language or a no-code interface. The goal is to turn robots from rigid, scripted machines into adaptable systems that can handle the messy, high-variability workflows that have long been the domain of human labor. It is a bet on making robots useful, not just precise, and it just secured a $32 million Series A from Engine Ventures to prove it [Morningstar, Feb 2026].
The wedge in material handling
Trener is not trying to build a better robot. Instead, it is building the software that makes existing robots from Universal Robots, Fanuc, or ABB smarter and easier to use. The initial wedge is material handling: the loading, unloading, and transferring of parts between industrial machines like injection molders, CNC mills, and grinders [Preqin, retrieved 2026].
This is a category of work that is notoriously difficult to automate with traditional methods. A fixed script works perfectly if every part is identical and arrives in the exact same position every time. In the real world, parts can shift on a conveyor, arrive in mixed batches, or have subtle dimensional variations. Programming a robot to handle that variability traditionally requires complex vision systems and even more complex code, pushing projects into the realm of being too expensive and fragile to justify. Trener’s argument is that an AI model trained on the physics of manipulation can adapt to that variability on the fly, turning a cost-center project into a viable one.
From code-defined to model-defined
The shift Trener is banking on is a fundamental one in industrial automation. Asad Tirmizi, the company’s CEO, frames it as a move from "code-defined workflows to model-defined skills" [AIPressRoom, May 2026]. Instead of a programmer writing explicit instructions for every possible scenario, an engineer would use Acteris to deploy a skill like "pick part from bin and place in fixture." The platform’s AI handles the perception, planning, and execution, adjusting its movements based on real-time sensor data. A digital twin allows for simulation and testing before anything is installed on the factory floor [Engine Ventures, retrieved 2026].
The promise is faster deployment and lower ongoing costs. A system that can be commissioned in weeks rather than months, and reconfigured by a plant engineer rather than an outside specialist, changes the unit economics of automation. It makes robotic automation accessible for mid-volume, high-mix production runs,the kind that make up a significant portion of discrete manufacturing.
A team built on robotic physics
The founders bring a deep, academic grounding in the problem. CEO Asad Tirmizi spent over 14 years working on AI and software frameworks for robotics before founding Trener [LinkedIn, retrieved 2026]. CTO Lars Tingelstad was previously an Associate Professor of Robotic Production at the Norwegian University of Science and Technology, where he focused on optimization and geometry for industrial applications [LinkedIn, retrieved 2026]. This isn’t a team applying large language models to a new surface; it’s a team that has been working on the core physics and control problems of robotic manipulation for years.
That pedigree has attracted a strategic investor syndicate. The Series A was led by Engine Ventures, with participation from Nikon’s NFocus Fund, Geodesic Capital, and Cadence, among others [Morningstar, Feb 2026]. Nikon’s involvement is particularly telling, as a company with deep expertise in precision optics and manufacturing seeing value in a software layer for machine vision and control.
| Investor | Round | Amount (USD) | Lead Role |
|---|---|---|---|
| Emergent Ventures | Seed (2024) | $5.4 million | Lead [Preqin, retrieved 2026] |
| Engine Ventures | Series A (2026) | $32 million | Lead [Morningstar, Feb 2026] |
| Nikon's NFocus Fund | Series A (2026) | Participated | Strategic [Nikon, Feb 2026] |
Where the wheels could come off
The ambition is clear, but the path is paved with technical and commercial challenges that have sunk previous attempts at flexible automation.
- The simulation-to-reality gap. Training AI skills requires vast amounts of data. While a digital twin helps, the physical world introduces friction, wear, and unpredictable lighting that can confound even well-trained models. Proving robustness across hundreds of unique factory environments is a monumental task.
- Integration complexity. Acteris is robot-agnostic, but factories are not. Retrofitting the software into existing workcells filled with legacy machines, sensors, and PLCs is where many automation projects grind to a halt. Trener is relying on a network of integration partners to standardize this process, which adds another layer of coordination [trener.ai, retrieved 2026].
- Proving the ROI. The ultimate customer is a plant manager with a capital budget. They will measure success in reduced labor costs, increased throughput, and minimized downtime. Trener must demonstrate that its AI-driven approach delivers a faster payback period than a traditional, fixed automation solution or the status quo of manual labor.
The competitive landscape is also formidable. Bright Machines has been building a software-defined approach to factory automation for years. Covariant has raised significant capital for its AI-powered robotic picking, primarily in logistics warehouses, a adjacent domain with similar technical challenges. Trener’s focus on machine tending within discrete manufacturing is a sharper wedge, but it will need to defend that territory as those well-funded competitors look for expansion.
The next twelve months
With the Series A capital, the next phase for Trener is about moving from promising technology to proven deployment. The key milestones to watch will be the announcement of named, referenceable customers in automotive, aerospace, or precision manufacturing. Public partnerships, like their recent webinar with Universal Robots, are a start, but paid deployments are the only metric that matters [trener.ai, retrieved 2026].
The company also needs to scale its library of skills. The initial focus on material handling is a logical beachhead, but the platform’s value grows with the breadth of tasks it can perform. Expanding into assembly, inspection, or finishing processes would signal both technical progress and market traction.
For a sense of the stakes, consider a back of envelope calculation. A skilled robot programmer might bill $150 an hour. Commissioning a complex machine-tending cell can take 200-400 hours of their time, a cost of $30,000 to $60,000 before any hardware is purchased. If Acteris can cut that commissioning time by 75% and allow a plant engineer to handle changes, the savings on a single cell pay for a multi-year software subscription many times over. The real prize, however, is unlocking automation for hundreds of tasks that were previously uneconomical.
Trener Robotics is not trying to beat the robot makers. It is trying to beat the inertia that keeps those robots parked in repetitive, simple tasks. Its real competition is the spreadsheet that shows the payback period is still too long, and the specialist whose flight has been delayed. If Acteris works as promised, the factory of the future won't be staffed by fewer people, but by people who can finally tell the robots what to do.
Sources
- [Morningstar, Feb 2026] Trener Robotics raises $32M Series A to bring Physical Intelligence to Industrial Automation | https://www.morningstar.com/news/business-wire/20260210315522/trener-robotics-raises-32m-series-a-to-bring-physical-intelligence-to-industrial-automation-providing-a-foundational-intelligence-layer-that-enables-software-defined-control-of-robots
- [Preqin, retrieved 2026] Trener Robotics profile | https://www.preqin.com/data/profile/asset/trener-robotics/717273
- [Engine Ventures, retrieved 2026] Trener Robotics | Engine Ventures | https://engineventures.com/companies/t-robotics
- [AIPressRoom, May 2026] Asad Tirmizi on Why Factories Are Moving from Programs to Models | https://aipressroom.com/asad-tirmizi-trener/
- [LinkedIn, retrieved 2026] Asad Tirmizi profile | https://www.linkedin.com/in/tirmizi/
- [LinkedIn, retrieved 2026] Lars Tingelstad profile | https://www.linkedin.com/in/lars-tingelstad-9b45674/
- [Nikon, Feb 2026] Nikon invests in Trener Robotics | https://www.nikon.com/company/news/2026/0213_01/
- [trener.ai, retrieved 2026] Trener Robotics website | https://trener.ai/