We started by building tools for our own work, then found other teams needed the same thing. That order matters: every product here solved our problem before it was ever sold.
We ship our own products and take on the engagements that make those products better. When the two conflict, the product wins — and we will say so before you sign anything.
These are in our contracts, not just on this page. If we break one, say so and we will fix it or refund the sprint.
Every tool on this site runs our own workload first. The gateway routed our internal traffic for four months before a single client touched it — which is why the failover path is the part we trust most.
No endpoint disappears without twelve months of notice and a migration path we write for you. If we ever cannot honour that, the source is yours to fork.
On the self-hosted tier we never see a payload at all. On managed, retention is set per field and deletion propagates to the index, not just the database.
Incidents get a written post-mortem within five working days, visible to every customer — including the ones that only affected one account.
A reply comes from someone who can read the trace and change the code, in business hours, Malaysia time. There is no tier-one script.
Export your routing config, your evaluation set and your full request history in one command. We would rather you go cleanly than stay because it is hard to unwind.
Founder · Engineering
Every project's accuracy bar goes through him. He has stopped more pilots than he has shipped.
Infrastructure
Spent six years on call for a payments platform. Designs for the bad Tuesday, not the demo.
Data
Will not start until someone can say, in one sentence, what a correct answer looks like.
Product · Front-end
Rejects any copy that reads like it was machine-translated, including ours.
Not the roadmap — the one integration that keeps breaking, or the invoice nobody can explain. That is where this usually begins.