AI

Undo lands $37M to give AI agents the runtime context to fix bugs

U.K.-based software debugging startup Undo Ltd. today revealed it has raised $37 million in new funding to expand internationally and push its debugging technology into the tools engineers use to fix code written with artificial intelligence.

Founded in 2005, the company records the full execution history of a running program and saves it to a single file. Most debugging tools show the code itself. Undo’s recordings capture what it actually did when it ran. The company sells the technology to engineering teams at networking, database, semiconductor and financial services firms.

AI coding assistants have made that distinction matter more. Teams are now shipping far more code than they can fully understand. Undo argues that an AI agent cannot reliably track down why such code fails unless it can see how the software behaved at runtime. Feed the agent that history, the company says, and it can pinpoint the cause of a bug rather than guess at it.

The company says its own benchmarks back the claim. On a set of complex bugs, the latest AI models identified the root cause 38% of the time on their own, a figure that rose to 92% when the models were given Undo’s runtime recordings, according to the company. Undo also says the recordings cut token usage on problems the models could already solve and that customers report completing root-cause analysis up to 100 times faster.

Palo Alto Networks Inc. is a customer. Suresh Sangiah, its senior vice president of engineering, said the costliest bugs in large codebases live in runtime states that logs do not capture. He said Undo often pinpoints the cause autonomously in minutes. “Undo provides the visibility needed to catch and correct errors before they become an operational problem for our customers,” he said.

The funding round was led by Elsewhere Partners, an Austin-based private equity firm that backs growth-stage software companies and has invested in 18 companies across North America, Europe and Israel since 2016.

The funding will go toward hiring on Undo’s product, support and sales teams in the United States and Europe.

Founder and Chief Executive Greg Law said the funding arrives as the technology becomes more relevant. “Undo has spent years building deterministic program recording technology for code failure runtime visibility, which has become absolutely essential with the rise of AI,” he said. “This investment allows us to accelerate at exactly the right moment, embedding Undo into AI workflows, scaling our commercial reach and ensuring we are an essential part of how engineering teams operate.”

Image: Undo