Services
Price quoted in writing in our first reply. You own the code.
What we do
A small decision model trained on your data that runs on ordinary CPUs, inside your agents and products. We build it, measure it and hand it over, the way we built jevx.
jevx →We prepare the training data and train your model on rented GPUs from prepaid credit (vast.ai, Kaggle), not a standing GPU bill.
Our sign-in product, customised and set up for your organisation.
Read →A senior engineer and our agent fleet ship AI features into your product every week. No lock-in; you own the code.
Your API as a CLI, agent skills and an MCP server, shipped in one week, with a forward-deployed engineer.
MCP servers over your internal APIs, databases and documents, with real authentication and per-tool permissions.
Production software through a gated agent fleet: spec, build in isolation, independent gate, verified commit.
Internal copilots, knowledge systems over documents, code and databases, and tool-enabled AI in production.
Read →Tool-calling agents, RAG over your own data, Spring AI integrations, agent skills for Claude Code and Codex.
Independent audits that turn risk into a prioritized roadmap with effort and ROI. Engineer to engineer.
Read →From one server to a resilient platform: CI/CD with safe rollbacks, OpenTelemetry, tested backups.
Read →Assessments of codebase, architecture, performance and security posture, with a prioritized roadmap.
Monthly engagements for live systems: scheduled fixes, or response terms agreed per critical system.
For our open-source tools: set up in your environment, supported in production, taught to your team.
AI employee
Our own inbox, last 7 days: 148 of 221 inbound emails filed by the agent with no human; the rest handed to a person.
Our own back office; per-email timing; the model was capped for part of the week.
Tell us the job →