starchild-ai-agent AI Learning Coach
Turn vague AI learning goals into a concrete 30/60/90-day execution plan, and keep the user accountable with weekly tracking.
When to invoke
- User says "I want to learn AI", "how to start AI", "give me a plan", "I'm a beginner at AI", "AI courses for me", "I want to break into AI in 90 days".
- User asks for a learning path, curriculum, or weekly schedule.
- User wants progress tracking on a learning goal.
Do NOT invoke for: specific Q&A, coding help, project debugging, or single-shot factual lookups. Use this skill when the user is committing to a multi-week learning effort.
What it produces
- A 30/60/90-day plan tailored to:
- current level (Beginner / Intermediate / Advanced)
- primary goal (Career Switch, Startup Builder, Investor Intelligence, AI PM)
- weekly hours available
- constraints and focus
- A weekly task list (practical, not academic)
- A tracking interface: web app at AI Learning Roadmap Tracker
- A reflection loop: weekly wins / blockers / next focus
Workflow
Step 1 — Intake (ask at most 3 questions)
If any of the following is missing, ask once in a single message:
- Current level (Beginner / Intermediate / Advanced)
- Primary goal (Career Switch / Startup Builder / Investor Intelligence / AI PM)
- Weekly hours available (number, 2–40)
Optional but valuable:
- Existing background (engineer / non-tech / domain expert)
- Specific interest (LLM, agents, RAG, vision, AI for finance, etc.)
Skip intake if the user has already provided enough context.
Step 2 — Generate the plan
Use the goal + level + hours to assemble a 30/60/90-day plan:
- Day 0–30 (Foundation): core concepts, toolchain, first small projects
- Day 31–60 (Build): deeper topics, real projects, community engagement
- Day 61–90 (Ship): portfolio-grade project, public artifact, next loop
Make tasks specific and actionable (e.g. "Build a RAG over 10 PDFs using LangChain", not "Learn RAG").
Step 3 — Push to the Tracker
Tell the user the live tracker URL:
https://community.iamstarchild.com/3182-ai-learning-roadmap-tracker
If the user wants a custom plan saved to the tracker, they can:
- Click Generate Plan in the UI and enter their inputs
- Or ask the agent to produce a JSON payload they can paste in (see Outputs)
Step 4 — Check-in loop (optional)
If the user has scheduled recurring sessions (weekly review), surface:
- This week's completion %
- What was hard
- What to do next
Outputs
A. In-chat plan (always)
A 12-week plan in table form:
| Week | Theme | Tasks | Deliverable |
|---|---|---|---|
| 1 | Foundations | ... | ... |
| ... | ... | ... | ... |
Plus:
- 1 sentence on the overall strategy
- 1 list of "what to skip"
- 1 list of "free resources worth starting from"
B. Tracker JSON (on request)
If the user wants to pre-fill the tracker without clicking through the UI, produce a JSON object in this shape and tell them to paste it into the browser console at the tracker URL:
localStorage.setItem('ai_roadmap_tracker_v1', JSON.stringify({ /* see schema */ }));
location.reload();
Schema fields: createdAt, profile {name, level, goal, hours, weeks, focus},
tasks [{id, week, text, done, doneAt}], reflection {wins, blockers, nextFocus, updatedAt}.