Undo gets $14M to scale to meet the software accountability challenge
Undo, a long time player in the debugging tools space, offering its program execution capture and replay technology to help others diagnose software failures, has closed a $14 million Series B round led by Cambridge Innovation Capital, the Cambridge, UK-based builder of tech and healthcare companies.
The 2005 founded startup — initially bootstrapped (out of founder Greg Law’s garden shed) — has come a long way, and now has more than 30 paying customers for what it describes as its “record, rewind and replay” debugging technology, including the likes of SAP HANA, Mentor Graphics, Cadence and Micro Focus.
A quick potted history: In 2012, Law quit his job to go full time on Undo, raising a small amount of angel funding and then a $1.25M from seed investment in 2014, followed by $3.3M in a series A funding in 2016.
New investors in the Series B round include Global Brain Corporation, a Japanese venture capital fund; and UK-focused Parkwalk Advisors, while all Undo’s existing investor groups also participated — including Rockspring; Martlet; Sir Peter Michael (founder of Quantel, Classic FM and California’s Peter Michael Winery); the Cambridge Angels group and Jaan Tallinn (co-founder of Skype and Kazaa).
The Series B will be used to expand Undo’s software development team, accelerate product development and grow its US operations. Undo says its best markets so far are electronic design automation (EDA); database manufacturers/data management; and networking.
“This funding will be used to significantly improve performance as part of Undo’s always-on recording vision, and also to accelerate our product roadmap and broaden the technology beyond compiled code so that it can be used with Java and other VM-based languages,” it tells us.
“Our main competitor is the status quo — engineering organisations that do not evolve with the times. Old-school debugging techniques (e.g. printf, logging, core dump analysis) have been around for decades. 2000 was all about static analysis. 2010 was about dynamic analysis, 2020 will be about capturing software failures ‘in the act’ through capture & replay technology.”
Undo argues that its Live Recorder technology offers “a completely new way of diagnosing software failures during development and in production” — arguing that its approach is superior to traditional debugging techniques such as printf, logging, core dump analysis which are “general purpose and provide limited information”, while it says static and dynamic analysis “are deep but can only look at specific instances of bugs” — whereas it claims its tech “can capture failure instances across the whole spectrum and therefore plugs in the gaps which no-one else has filled yet”.
The UK company also sees a growing opportunity for its approach given increasingly complex and increasingly autonomous software risks becoming unaccountable, if it’s making decisions without people knowing how and why. So the wider vision for Undo is not just getting faster at fixing bugs but addressing the growing need for software makers to be able to articulate — and account for — what their programs are doing at any given moment.
“Longer term it’s about that journey towards software accountability,” says Law. “Software accountability is quite a broad thing — it really means the ability to be able to know for sure what some software did as it ran. And today that’s all about the programmer’s understanding of what their program has done. But actually it’s far more than just programmers that need to understand software — and particularly as we move into this second chapter of the information revolution where computers are beginning to make decisions that affect our lives and our livelihoods. I mean in the case of social media and Facebook and things, Western democracy! The ability to have that accountability behind software actions is going to become a really important thing. That’s a progressive journey that we’re on.
“So the question is what did the software actually do? And as we grow, and as time goes on, we’ll answer that question in progressively bigger and bigger contexts.”