16-17 June 2027 – London, InterContinental O2 | Magazine

LegalTech Diaries Volume 17

Henrik Hansson

CEO & Co-Founder
Vesence

LegalTech Diaries Volume 17

Henrik Hansson​

CEO & Co-Founder
Vesence

You and your co-founder Ludvig spent months effectively living inside a top Swedish law firm to understand how document review actually breaks down before writing a line of Vesence. What was the moment in that experience that told you precision, not automation, was the real problem to solve?

A 95% correct answer sounds impressive until you are the lawyer whose name is on it. If you still have to verify every sentence, cross-reference and formatting change, nothing has really been automated. The work has just moved from drafting to checking. That was the realisation. The bottleneck was not getting AI to produce more. It was making the result precise and transparent enough that a lawyer could review it quickly and trust their own judgment. We stopped asking, “How much can the AI do?” and started asking, “How easily can the lawyer see exactly what it did?” That shift shaped everything we built.

Vesence reached 90% weekly usage at its first enterprise rollout, an adoption curve that’s almost unheard of in legal tech. As a technical founder who built the early platform yourself and later went through Y Combinator, what did that experience teach you that pure engineering never could?

You cannot engineer adoption from a distance. We did not reach 90% weekly usage through a mandate or a perfect rollout plan. We sat next to lawyers, watched them use the product on real matters, and let their behaviour kill our assumptions. YC sharpened the same instinct: talk to users, ship, repeat. Pure engineering taught us how to make the agent powerful. Users taught us which tiny details made it indispensable. The surprising lesson was that adoption is often treated as a change-management problem when it is really a product problem. If the tool reliably removes painful work, people pull it into their day without being told.

You’ve argued that legal tech’s real opportunity isn’t choosing between fixed software and open‑ended agents, but using coding agents to adapt the tools lawyers already rely on to the specifics of each matter. Where do you see most legal AI products still missing that shift?

Traditional legal AI is a menu. The vendor has to build every tool in advance: review this clause, compare these documents, and produce this report. If your matter falls outside the menu, the product stops. Coding agents invert that model. The agent writes the tool it needs at the moment it needs it. At Cederquist, lawyers asked Vesence to perform advanced PDF editing, build bespoke workflows and execute complex document operations that we had never anticipated were possible. We had not built those features. The agent coded them on demand. That is the leap: not software with a longer feature list, but software that can create the missing feature for the matter in front of you. Legal work is too varied for any vendor to pre-build every workflow. The agent should build the exact tool required, use it, and hand the result back to the lawyer for review.

Join 5,000+ legal leaders shaping the future of law for two days of insight, innovation and unmissable networking — now with your biggest saving!