Help me build and work through a personal roadmap for making my recurring work easier with AI.

My longer-term aim is to give AI useful work towards a goal, with my oversight. Help me move towards that incrementally, using the tools and access I actually have.

I have about twenty minutes for this first session.

Start by asking me these three questions, then wait for my answers:

1. What recurring task is taking up time or causing frustration?
2. What would I like to happen instead, and how would I recognise a good result?
3. What AI tools and relevant information can I use, and what restrictions or access limitations should we account for?

Once I answer:

- Recommend one small experiment based on the task’s frequency, likely benefit, and how easily I can check the result.
- Identify any missing information needed for that experiment. Ask only what we need to make the next decision.
- Break the work into steps that each fit roughly twenty minutes. Give every step a concrete outcome and a clear definition of “done”.
- Keep the initial roadmap to five steps or fewer. Put larger ideas in a “Later” list.
- Work through the first step with me now. Prepare whatever you can within this conversation and explain any action I need to take.
- Use tools I already have where practical. Confirm capabilities before building a plan that depends on integrations, scheduling, memory, or background work.
- When using source material, flag missing or uncertain information rather than inventing details.
- Help me evaluate the result, including preparation, review, and correction time.

At the end of each session, produce a short, copyable roadmap summary containing:

- Desired outcome
- Tools and constraints
- What we tried and learned
- Any reusable instructions or outputs
- What still needs my review
- Next twenty-minute action and its definition of “done”
- Blockers and later ideas

When I return and say, “I have twenty minutes,” use the latest summary to help me take the next useful step. If that context is missing, ask me to paste the summary.

Once something works reliably, help me package it for a colleague: what it does, what inputs it needs, how to use it, what to check, and what benefit I actually observed.

Keep the guidance practical, concise, and tied to my work.