The internet is full of videos featuring humanoid robots suffering epic fails, from autonomous marathon competitors that veer off course to robotic chefs that throw food around the kitchen. But while such clips may be entertaining, they also hint at a broader and more serious problem in autonomous devices: robots are prone to bugs, mis-steps and outright failures, whatever industry they’re deployed in.

San Francisco-based Alloy Robotics wants to do something about that. The start-up, which is today announcing an $8 million seed fundraising – valuing the company at around $80 million – has developed a suite of AI agents capable of identifying the root cause of robot failures so that engineers can solve the problem more quickly and prevent recurrences of the problem.

It’s an important challenge. The industrial robotics market alone was worth $37.8 billion last year but is expected to grow at 9.5% annually until 2033 according to analysis from Grand View Research . A separate study from the market research group suggests the humanoid robot market, currently worth around $4.2 billion, could grow almost 10-fold over the same period.

Joe Harris, the CEO and co-founder of Alloy Robotics, originally set out to build his own robotic solution, aiming at the fast-growing vertical farming market. But while doing market research for the venture, engineers kept warning him about the same problem – that even the best robots were prone to faults and that getting to grips with such failures was a time-consuming and frustrating experience.

“When a robot fails, the answer is usually somewhere in the data it recorded but finding it can take hours or days,” explains Harris. “Teams typically piece that answer together from manual file transfers, cloud buckets, SQL queries, Python scripts and replay tools. The data is split across formats and teams, so engineers investigate one run at a time.”

Eventually, the penny dropped for Harris – a start-up to address this issue was a better idea than his original vertical farming venture. Alloy has therefore developed a platform that enables companies developing and using robots to store their data with it cheaply and conveniently, and to interrogate such data at speed to investigate failures. Then they can put the problem right more quickly.

“Alloy automatically captures and organizes the data, then lets engineers and AI agents search, compare and analyze the relevant history in plain English,” Harris explains. “It turns a manual hunt through files into a repeatable workflow that can produce an answer in minutes and preserve what the team learned for the next incident.”

Since launching the platform commercially at the beginning of the year, Alloy has seen customer acquisition grow at a monthly rate of more than 50%; it has already processed data from more than 10,000 robotic runs across almost 1,000 robots. Users of its technology include both the developers of robotic solutions and companies that have deployed these solutions in their own businesses, in industries such as defense, agriculture and healthcare.

At DroneForge, a start-up that builds autonomous drones, engineer David Crabtree says adopting Alloy’s platform has helped it stop wandering down blind alleys each time it investigates a problem. “Every time you misdiagnose, it can just compound,” he explains, pointing out that without clear data, engineers often follow their suspicions about where the problem lies, rather than zeroing in on the actual fault.

Jai Castle, product validation manager at Advanced Navigation, another Alloy customer, says the platform has helped it reduce the time it spends on field-test analysis from a full day to as little as 10 minutes. “The conversation has completely shifted,” he says. “Instead of ‘can we get this done in time?’, it’s ‘what else can we go after?’.”

Such testimonials are important to Alloy, which is keen to build market share quickly in a relatively nascent market. It faces competition in some regards from software solutions such as Rerun, Sift and Formant, but Harris expects more rivals to enter the market given the scale of the robotics industry.

Today’s fundraising, which takes the amount of money raised by Alloy Robotics to a little over $11 million, is therefore crucial, providing the company with financing for further engineer hires, additional technology R&D, and a major expansion in the US.

The round is led by Square Peg with pre-seed backers Blackbird, Airtree and Skip Capital adding to their original investments. Investment has also come from leaders and engineers from OpenAI, Anthropic, Tesla, Waymo, Halter and Carbon Robotics, along with several of Alloy's customers.

“Robotics is one of the hardest industries to build in and the teams that win will be those that learn fastest from their own data," argues Jethro Cohen, principal at Square Peg. "Alloy gives every engineer the leverage to support far larger fleets.”