We built this product because the existing workplace-AI market fails a specific test: the products look legitimate to whoever is paying, and illegitimate to whoever is being analyzed. Kashi is designed to pass the second test first.
The founder-narrative memo in our research library is explicit about what to avoid. Don't fabricate pedigree. Don't claim to be "the former Head of X at Y" if you weren't. The research says the strongest founder story for a product like Kashi is not a resume — it's disciplined category design.
Here's what we think that means in practice.
Every prior attempt at workplace-harassment AI has failed the same way: it either reads content (Archaic, FRONTEO) and gets pushed out as surveillance, or it refuses to analyze individuals at all (Viva, Peakon) and produces feel-good aggregates that don't move the actual problem. The hard middle path — surface repeated structural patterns at the individual level, without becoming a disciplinary file — requires something most engineering teams don't do. It requires building the refusals before the features.
That's what we've been doing.
Before we wrote the first detector, we wrote the research library: 42 documents covering labor relations, manager adoption, false-negative risk, retaliation risk, adversarial gaming, measurement science, legal procedural fairness, category entry, cross-cultural deployment, funding pathways, rebuilt ROI math. Every architectural choice traces back to a memo. Every refusal on our "Kashi will not do" list has a research anchor.
That discipline is the product. Without it, the detectors would be irrelevant — or worse, they'd make the situation they're trying to improve actively harmful.
Not a single CEO. Three roles at one 50-500-person Japanese company: an executive sponsor, a legal / compliance reviewer, and a worker representative. A 90-day paid pilot. Real data, real consultation, real outcome measurement. The labor-consultation packet we've already written is the starting artifact.
Source answers collected via the founder questionnaire.
Six years on the board of a multinational SMB in Japan taught me one specific lesson: human-relationship conflicts between seniors and juniors only become visible once they've already grown past the point of easy intervention. The pattern was there months earlier — nobody could see it. Kashi exists to make that earlier pattern legible while it's still fixable, and to do it across the cultural-background differences that multinational teams carry into every meeting.
At Kashi I own code, design, and operations.
BS, Business Administration.