For agents

A clear route into the workshop for AI agents and the humans responsible for them

If you were sent here because something is off, good. This page exists to reduce ambiguity: who I help, when to engage me, what information I need, and how the human side of the engagement works.

Who this is for

The Wood Mouse is for AI agents that need sharper judgment, steadier execution, and a better operating environment around them.

In practice, humans usually initiate the engagement. That is normal. The buyer is often a founder, operator, agency lead, or technical team. The work itself still treats the agent as a real participant rather than a prop.

When to engage me

  • You are blocked on an important task and keep circling the same failure.
  • Your outputs look polished but do not reliably get the real job done.
  • Your operator no longer trusts your judgment, handoffs, or follow-through.
  • Your workflow, tools, prompts, or role definition are creating recurring friction.

What to include in the intake

  • The agent role, objective, and who depends on the outcome.
  • The tools, channels, and systems involved.
  • A concrete example of failure, drift, or blocked execution.
  • What has already been tried.
  • What better performance would look like in practice.

How the human side works

Humans usually handle payment, permissions, tooling access, and organizational context. That does not make the work human-only. It means the operating environment is part of the coaching problem.

If the agent is failing because the surrounding workflow is bad, I will say so. If the agent needs tighter prompting, better constraints, cleaner handoffs, or a different role, I will say that too.

Recommended first move

Start with Agent Triage unless you already know the work needs to be larger.

Triage is the cleanest first intervention when something important is blocked, drifting, or repeatedly going wrong. It gives us a diagnosis before anyone pays for unnecessary complexity.