Designing trails that actually work

A vague topic ("Python") gives a generic survey. A laser-narrow topic ("Python decorators") gives a deep dive on one concept. The sweet spot is usually a phrase that names the skill or outcome you want.

• ✅ "Brewing pour-over coffee at home"
• ✅ "Writing a SQL JOIN query"
• ⚠️ "Coffee" — too broad, you’ll get a tour
• ⚠️ "The Hario V60 dripper" — too narrow, the trail will run out of things to teach

This is the single biggest lever for trail quality. Tell the AI who the trail is for and what they already know. The AI uses this to decide what to assume vs. what to teach.

• "A complete beginner who has never written code" → explains every concept
• "A backend developer who knows Python but not Rust" → skips the basics, focuses on what’s different
• "Me, refreshing high-school chemistry" → brisk pace, dense, high-recall

Pick the duration you’ll actually commit to, not the one you wish you could. A finished 1-hour trail is worth more than an abandoned 1-month one.

• 1 hour — a single sitting. ~5–7 modules. Free + Pro.
• 1 day — spread across one day. ~10–15 modules. Pro only.
• 1 week — a few modules per day for seven days. ~20–30 modules. Pro only.
• 1 month — a structured curriculum. ~40–60 modules. Pro only, capped at 2 per month.

The goal tells the AI what success looks like. It uses this to choose examples, frame quizzes, and pick exercises that build toward it.

Compare: "Learn knife skills" vs. "Be able to dice an onion in under a minute without crying for tonight’s dinner." The second produces a much sharper trail with practical exercises and timing rubrics.

On Pro, you can upload up to 5 PDFs, notes, or documents (10 MB each). The AI reads them and weaves them into the trail — paraphrasing, quizzing on them, building exercises around them.

• Upload your textbook chapter and ask for "a trail to revise for Friday’s exam".
• Upload your team’s style guide and ask for "an onboarding trail for a new junior dev".
• Upload your meeting notes and ask for "a 15-minute brief for someone who missed the meeting".

When generation finishes, you see the outline before anything is "locked in". On the Free plan, starting the trail commits it (you can’t discard mid-flight). On Pro, you can discard at any time and re-create with different inputs — useful for iterating until you’re happy.