Make Your Product Agent-Ready
Agents are becoming a way software gets used. Most products cannot be driven by one: no documented API, no machine credentials, nothing an agent can discover on its own. These workshops fix that, and teach your engineers to build with Claude Code properly.
Why this training is different
Taught from a published standard
The agent-readiness material is built on Wired for Agents, a four-criterion standard I wrote and published. You are not being taught one consultant's opinion. You are being taught against criteria you can read, argue with, and verify yourself.
Your codebase, not a sandbox
Exercises run against your own repository. Teams leave with work committed, not with notes about work they might do later.
Anthropic only, deliberately
Claude Code and the Claude API, in depth. Tool-agnostic training teaches the shallow overlap between products. Going deep on one is what changes how a team works.
Taught by someone still shipping
Seventeen years building software, currently writing agentic systems in production. Not a researcher, not an evangelist.
Who this is for
Built for engineering teams
These are workshops for people who own a codebase. If you are looking for an introduction to AI assistants for general business use, one-to-one tutoring is the better fit.
Engineering leaders
- Deciding whether agent support is worth building
- Watching engineers use Claude Code inconsistently
- Needing a standard the team can be held to
Platform and API teams
- Owning the surface agents would call
- Weighing MCP against the existing REST API
- Stuck on auth that assumes a human is present
Product teams
- Asking what agent-ready would actually mean
- Trying to scope the work honestly
- Wanting to know what it buys before committing
The workshops
Two workshops, run privately for your team
Both run on site or remotely. Delivered for a single team rather than as a public course, so the material and the exercises use your systems.
Make Your Product Agent-Ready
Half day or full day, for engineering and platform teams
The four criteria
- A documented, versioned API an agent can read (WFA-1)
- An MCP server whose tools match what the API does (WFA-2)
- Scoped, revocable machine credentials with no human in the loop (WFA-3)
- Discovery from a bare domain, before authentication (WFA-4)
What the team does
- Audits your product against all four criteria
- Writes or corrects an OpenAPI document for a real endpoint
- Stands up a minimal MCP server against that endpoint
- Works through the auth change, usually the hardest of the four
What you leave with
- A written gap analysis against the standard
- A sequenced plan for the gaps you chose not to close that day
Claude Code for Engineering Teams
Half day or full day, for teams already using it unevenly
What it covers
- Giving Claude Code the project context it needs to be accurate
- Splitting work so an agent can finish it, and recognising work it cannot
- Reviewing agent-written code, including what reviewers reliably miss
- Where agents genuinely should not be used
What the team does
- Runs a real task in your repository, start to merged
- Reviews agent output against your existing standards
- Drafts the conventions file the team will actually keep
What you leave with
- A written convention for how your team uses agents
- A review checklist tuned to how agent code fails
Practicalities
How a workshop runs
Half day or full day
Half a day covers the material and one worked exercise. A full day gets meaningful work committed. Teams new to the topic should take the full day.
Up to a dozen engineers
Small enough that everyone gets their own work reviewed. Beyond that the exercises stop being hands-on.
On site or remote
On site across Virginia and the DC metro, including Roanoke and Richmond. Remote anywhere, run as two shorter sessions rather than one long one.
$2,500 half day, $4,500 full day
One flat fee for the team, not per seat. On-site travel outside the DC metro is billed at cost. Both workshops booked together are $8,000.
Jason Flatford
Founder, Mythic Works
Seventeen years building and shipping software. I write agentic systems in production and use Claude Code daily on real codebases, which is where this material comes from.
I wrote and published Wired for Agents, the standard the first workshop teaches against. I am not an AI researcher and not an evangelist.
How I use these tools in my own client work, including what they are allowed near and what happens to client material, is written down in the AI use policy. I teach against the same standard I hold myself to.
- 17+ years shipping software
- Built products serving 400k+ users
- Author of the Wired for Agents standard
Talk it through first
Fifteen minutes is usually enough to work out which workshop fits, or whether either does. If neither is right I will say so.