The hard part is building the infrastructure around the work well enough that it survives contact with the real world.
I learned that early. As a law student at Kenyatta University School of Law, I started Kenyan Legal — a student-authored law magazine, almost no budget, distributed across 19 Kenyan law universities. Then Africa Law Times: a pan-African legal journal with contributors from Uganda, Nigeria, and across East Africa, covering the questions that actually mattered — Africa's relationship with the ICC, the fragility of new democracies, the gap between constitutional promise and legal reality on the ground. A proper editorial board. A serious publication.
What those years gave me was not legal training. It was a precise understanding of how Kenyan advocates find and use legal information — slowly, imprecisely, and with tools that were never built for this market.
When AI research tools began producing fabricated citations that reached court filings — and advocates began facing professional consequences for it — I recognised the problem immediately. Not because I read about it. Because I had spent years watching exactly how that gap between a research tool and a submission-ready output gets bridged in practice: usually by a junior, under time pressure, with no verification layer between the output and the filing.
Jurislexr is a product of Kontorva Oy, a Finnish–Kenyan company that puts that infrastructure to work — citation-verified research, a tamper-evident verification stamp, Kenya-region data residency — built for work product that carries professional consequences.
This is not a tool for moving faster and worrying about accuracy later. It is for practitioners who understand that what leaves their office carries their name.
