Services
Three ways to work with us, for teams putting AI into real decisions: we build a decision system with you, review the agents you already run, or teach the practices behind our products.
Build a Decision System
What decision does your team repeat? We map it, wire research + challenge + evidence, ship the workflow, then measure outcomes — the same loop behind Invest, Travel, and Cook.
- Map one recurring decision your team already makes
- Research, challenge, and evidence in one production flow
- Ship, measure what changed, then tighten the loop
Agent Security Assessment
A 1–2 week, read-only review of your AI agents, MCP servers and tool permissions. You get a findings report ranked by risk, a permission manifest for every agent, and a remediation walkthrough.
Decision Workshop
A half-day or full-day session, remote by default: map one recurring decision, then walk research → challenge → evidence → action on your stack.
Topics to choose from
- Writing decisions down: architecture decision records that state the context, the tradeoffs and the limits
- Evaluating LLM features: datasets, metrics and regression gates that can block a bad release
- Taking MCP to production: gateways, permission manifests and approval gates
- One answer, computed in one place: worker-writes / API-reads for AI products
What you get
- An agenda written for your team after a 30-minute call
- The session itself, using your own systems as the examples
- Written notes and a checklist your team can keep using
The assessment and the workshop are fixed scope and fixed fee, quoted after a short call. A custom system is scoped with you first. No retainer.
Something else?
If you have a specific agent, MCP surface or decision path in mind, describe it and we'll tell you honestly whether we can help.
