What we believe about AI

AI is not a replacement for engineering fundamentals. We treat it as leverage for them.

Two rules we work by.

  • Rules first. Write the rules before you write the prompt. AI builds what you describe, not what you forgot to describe.
  • Build to survive. AI tools, by default, optimise for working. Not for surviving. Security, testing, scaling and monitoring are built in, not added later.

This site works for AI agents as well as people.

  • llms.txtA short, structured guide to who we are, our services, case studies and how to reach us.
  • llms-full.txtThe full text of every service and case study, plus all the Insights articles.
  • A public MCP serverRead-only, no sign-in, with six tools covering services, 33 case studies, search and contact details.
  • Agent-ready systemsInterfaces we build on clients’ existing systems so AI assistants can handle tasks safely.

The MCP server.

Streamable HTTP at /api/mcp, JSON-RPC 2.0 over POST, no authentication, no sessions. All tools are read-only and safe to repeat; nothing an agent sends is stored.

Example requestJSON-RPC 2.0
POST /api/mcp
Content-Type: application/json

{"jsonrpc":"2.0","id":1,"method":"tools/call",
 "params":{"name":"get_service","arguments":{"slug":"systems-teardown"}}}

MCP tools.

  • list_servicesThe 9 services, each with a slug and a one-line summary.
  • get_serviceOne service in full, as Markdown. Argument: slug.
  • list_case_studiesAll 33 case studies with client, industry and outcome-led title.
  • get_case_studyOne case study: the business problem, what we did and what changed. Argument: slug.
  • search_contentSearch services and case studies. Argument: query.
  • get_contact_infoEmail, phone, offices and the right next step for each kind of visitor.