Siti Muinah
In an era where Artificial Intelligence is rapidly augmenting the software development lifecycle, the sheer velocity of code generation has created a paradoxical challenge: as engineers build faster, the systems they inhabit are becoming increasingly opaque. Cambridge-based scale-up Undo has emerged as a critical vanguard in this landscape, announcing today the successful close of a €31 million ($37 million) funding round. Led by Elsewhere Partners, this injection of capital is earmarked to propel Undo’s deterministic, program-recording technology into the mainstream of AI-integrated enterprise workflows.
As businesses pivot toward AI-assisted coding, the industry is witnessing a "black box" phenomenon. While AI agents are adept at churning out lines of code, the resulting systems often lack the transparency required for mission-critical reliability. Undo is positioning itself as the bridge between AI productivity and system stability, providing the "runtime context" that allows both human engineers and autonomous agents to understand exactly why code fails.
The Core Solution: Turning Runtime Chaos into Data
Founded in 2012, Undo has spent over a decade perfecting its core technology: a deterministic, program-recording engine that provides absolute visibility into code failure. Unlike traditional logging, which captures only what a programmer thinks might be useful, Undo records the complete execution history of a program.
This recording—a self-contained "flight recorder" for software—captures every state change, memory access, and branch decision. When a bug occurs, engineers do not need to guess; they can simply "rewind" the software to the exact moment of failure. In the age of AI, this is not just a debugging convenience—it is a survival mechanism. By feeding this granular runtime context into Large Language Models (LLMs), Undo allows AI agents to act as expert investigators, diagnosing complex race conditions and memory leaks that would otherwise take human teams weeks to resolve.
A Strategic Timeline: From Niche Tool to Essential Infrastructure
The trajectory of Undo mirrors the broader evolution of the software industry, moving from manual debugging to automated, AI-driven diagnostics.
- 2012 – The Inception: Undo is founded in Cambridge, UK, focused on the fundamental problem of software "observability." The company builds the foundation for record-and-replay debugging.
- 2015–2020 – Enterprise Adoption: The company gains traction among high-stakes engineering environments, including cybersecurity giants like Palo Alto Networks, where "guesswork" is not an option.
- 2023–2025 – The Generative AI Explosion: As organizations begin deploying AI-assisted coding tools, they report a sharp increase in "unmanageable" codebases. Undo pivots to emphasize its role as an AI-enabler.
- 2026 – Scaling for the Future: With the close of the €31 million round led by Elsewhere Partners, the company initiates a major expansion of its go-to-market teams across the US and Europe to meet the surging demand for reliable AI-first engineering tools.
The "AI-First" Market Context
Undo’s latest funding round is part of a broader, well-documented trend in 2026. The venture capital market has shown a clear preference for companies that address the "operational layer" of AI—the infrastructure that keeps the lights on while autonomous agents do the heavy lifting.
The UK tech ecosystem, in particular, has become a hotbed for this specific sub-sector. The rise of companies like Cloudsmith (supply chain security), Geordie AI (AI agent governance), and Overmind (agent security) signals a shift in investor sentiment. Capital is no longer just flowing toward "model builders," but toward the guardrails, observability tools, and security layers that make those models viable for the enterprise.
For Undo, the value proposition is simple: AI = Model + Context.
Without high-fidelity runtime data, an AI agent is essentially guessing at the state of a program. With Undo, the model is provided with the ground truth of the system’s behavior. This synergy is proving to be a decisive factor for enterprise CTOs who are wary of the instability often introduced by AI-generated code.
Official Perspectives: The Trust Gap
The consensus among industry leaders is that the current rate of AI-assisted code generation is outpacing our ability to verify that code.
Greg Law, Founder and CEO of Undo, emphasizes the urgency of the moment:
"We are ahead of the curve. Undo has spent years building deterministic, program recording technology for code failure runtime visibility, which has become absolutely essential with the rise of AI. 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 in this new, AI-first world."
Rod Favaron, Operating Partner at Elsewhere Partners, highlights the risks of ignoring the "black box" nature of modern software:
"AI is making code unmanageable—introducing code that engineers cannot understand, trust, or debug. So while AI helps them generate more code, some of it is poorly understood, poorly structured, and of questionable quality. Systems become full of unknowns, making them unstable and increasing the risk of outages, security breaches, and customer escalations. Undo ensures engineering teams can effectively operate complex systems in an AI-first world and provide the essential runtime context for enterprise-grade, AI-assisted software."
Supporting Evidence: The Palo Alto Networks Case Study
The true test of any developer tool is its utility in a massive, multi-million-line codebase. Suresh Sangiah, Senior VP of Engineering at Palo Alto Networks, provides a testament to why tools like Undo are no longer optional for large-scale enterprises:
"Quality is extremely important at Palo Alto Networks, and we cannot afford to rely on guesswork. The hardest—and costliest—bugs in multi-million-line codebases live in runtime state and are not captured by logs or other solutions. Undo provides the visibility needed to catch and correct errors before they become an operational problem for our customers, enabling automatic root cause analysis. Undo often autonomously finds the root causes in minutes."
This efficiency—reducing the time-to-resolution from days to minutes—is the primary driver of Undo’s ROI for its customers. By automating the "investigation" phase of a bug report, Undo frees up senior engineers to focus on architecture rather than incident response.
Implications: The New Era of Software Engineering
The closing of this funding round for Undo signifies a maturation of the AI software market. We are moving out of the "hype phase" and into the "infrastructure phase."
1. The Death of the "Debug and Pray" Workflow
Historically, software debugging has been a process of trial and error—reproducing a bug, adding logs, and hoping the next attempt captures the elusive state of the program. Undo’s technology renders this archaic. By capturing the complete execution history, the industry is moving toward a deterministic future where failures can be analyzed with the same precision as a video replay.
2. The Rise of the "AI-in-the-Loop"
The future of engineering will not be "AI vs. Human," but rather "AI-in-the-loop." In this workflow, the AI generates the code, the CI/CD pipeline tests it, and, when failures occur, the AI uses tools like Undo to inspect its own work. This creates a self-healing loop that is essential for scaling complex systems.
3. Institutionalizing Reliability
As regulations surrounding software liability and cybersecurity tighten, companies will be held more accountable for the stability of their AI-generated software. Investing in tools that provide observability and auditability is becoming a standard compliance measure. Undo is perfectly positioned to capture this demand as enterprises transition from experimental AI use cases to core-business production.
Conclusion: Looking Ahead
With a war chest of €31 million, Undo is set to scale its operations globally, bringing its "flight recorder" technology to a broader segment of the Fortune 500. As the company expands its footprint in the United States and across Europe, the challenge will be scaling its support and integration services to match the rapid adoption of AI development agents.
For the wider software industry, Undo’s success is a signal that while the excitement of AI generation is palpable, the hard work of building sustainable, debuggable, and reliable systems remains the bedrock of success. As we head further into 2026 and beyond, the companies that prioritize "runtime visibility" will be the ones that survive the transition to an AI-first world, while those that rely on guesswork may find themselves struggling to maintain the very systems they built to innovate.
Undo has proven that in the complex world of software, the most important step forward is often the ability to look back at exactly what happened.
Evan Lee Salim
Ali Ikhwan
