Create a Personalized Learning Roadmap from a Website with Runner H

What you’ll get

A personalized, week-by-week study plan built automatically from any website’s content. You point an AI agent (Runner H, a tool that can browse and act on its own) at a site, answer a few questions about yourself, and approve each week as it goes. The result is a Google Doc roadmap that fits your level and how much time you actually have. No coding; you just chat with the agent.

Tools you need

• Runner H (paid): an autonomous AI agent (one that can browse websites and carry out multi-step tasks). It reads the site and builds the plan.
• Google Docs (free): where the finished roadmap is delivered.

Steps

1. Give the agent its role: start Runner H and paste this instruction, which tells it to act as a learning-path architect that always waits for your approval:

   You are my Autonomous Learning Path Architect. Your mission is to build a fully
   customized, iterative learning roadmap from a website I give you. Never proceed to
   the next step without my explicit approval.

   You should see it accept the role and ask for the next input.

2. Give your email: it asks where to send the finished roadmap. Reply with your email address. It confirms and asks for the website next.

3. Give the website URL: paste the link to the site you want to learn from. The agent crawls it and pulls out the table of contents and sections.

4. Let it sort topics by difficulty: it tags each topic Beginner, Intermediate, or Advanced, so the plan builds up sensibly. You should see it confirm the topics are tagged.

5. Give your learning profile: it asks about you so it can tailor the plan. Answer each part, for example:

   Experience level: 3 out of 5
   Strong areas: Python, basic algorithms
   Challenge zones: advanced data structures, system design
   Availability: 60 to 90 minutes per day

   It then works out how many weeks you need and the hours per week.

6. Review and approve Week 1: it presents the first week (up to about 10 topics, each with a description, a time estimate, and a link). Approve it, or ask for changes:

   Go Week 2

   (Or give specific feedback to adjust it first.)

7. Approve each following week: keep reviewing and approving Week 2, Week 3, and so on. It builds the plan one week at a time, pausing for your okay each time, which is what keeps it tailored to you.

8. Finalize: when you are happy with every week, tell it to wrap up:

   Finalize Roadmap

   It generates a Google Doc titled “My [Site] Mastery Roadmap” and emails you the link.

Original source

Based on a DEV Community submission by dev_99718084404de for the Runner H “AI Agent Prompting” Challenge, describing a “Roadmap Architect” that turns a website into a personalized learning journey.

Notes & variations

• Free-tier alternative: Runner H is paid. You can imitate the idea for free with ChatGPT or Claude, but you would paste the website content in yourself and ask the AI to structure it, which is more manual.
• Common pitfall: vague answers in your learning profile (step 5). The more honest and specific you are about your level and time, the better the plan fits.
• Tip for better results: use the approval gates. If a week is off, give precise feedback before approving; that back-and-forth is what makes the roadmap genuinely yours.