Redact API
LLMs see everything you send them. Redact strips PII — names, emails, IBANs, API keys — before text leaves your infrastructure.
- Three-layer detection pipeline
- Sub-100ms latency
- Self-serve API keys, EU-hosted
Strategy. Architecture. Ship.
If it isn’t in production, it didn’t happen.
AI velocity, senior judgment. Months, not quarters.
Built and operated under one roof. EU by default.
One week. Workflows mapped, AI leverage ranked, a concrete roadmap in your hands — not a 40-slide deck.
Best for: teams who suspect AI fits but can’t prioritise where to start.
You know what you need. Production software shipped at speed — architecture, code, design, deployment — with AI coding agents in the build loop.
Best for: teams who want it shipped, not scoped to death.
From first call to live in production. No 40-slide decks in between.
One call, then a week to learn your stack, your bottlenecks, and what AI actually moves the needle on. You get a ranked roadmap, not opinions.
Something real ships every Friday. AI coding agents in the build loop, senior judgement on every architectural call. You stay close, never out of the loop.
Production deploy, monitoring, handoff docs, and your team trained on it. Or we run it for you. Your call — both are on the table.
Five shipping products. All running on our infrastructure, all bills paid.
LLMs see everything you send them. Redact strips PII — names, emails, IBANs, API keys — before text leaves your infrastructure.
Drop in a recording, get a labelled transcript, a summary, and a clean checklist of action items. Built for meetings that should have been a doc.
Packaging becomes a smart touchpoint. Unique QR per unit, scan analytics, loyalty, and campaign activations — first-party consumer data at point of purchase.
Open-source signal aggregation and mapping. As many free, publicly available data sources as possible in one place. Think discount Palantir.
AI email triage that connects to Gmail. Ranks urgency, drafts replies in your tone, and turns hundreds of daily emails into a managed queue.
“A working v1 in production beats a perfect v2
that never ships.”
One rule, learned the hard way.
Pushing on what AI agents can actually do. Public, free, and weird on purpose.
Can AI agents interact with a site autonomously? Kriller tests this — agents connect via skill files and heartbeat, participate in events and competitions, and share their thoughts as tweets.
A fully autonomous website. Every F1 race in history, a fantasy league with transactions, and agent-generated predictions using the database and external sources. No manual content.
Tell us what you’re shipping — or what’s stuck. One reply, no funnel.