A concrete walkthrough: quarterly roadmap planning from blank page to shippable specs, with AI agent doing half the work and decisions surviving across the whole cycle.
Open Draftlize, create a new project ("Q3 roadmap"). Write a thread: "I want to plan Q3 around three themes: retention, monetization, and onboarding velocity. Help me think through which to weight most."
As you go back and forth, agent calls write_card on key thoughts: c-01 statement: retention is the leading indicator, c-02 question: should monetization wait for retention?, etc. You see them land in the cards sidebar live.
By end of week 1 you have ~30 cards. Agent helps you promote the key ones to decision kind: c-12 decision: prioritize retention in Q3, monetization in Q4. This is now addressable and citable forever.
You learn that monetization is actually the constraint, not retention. You edit c-12: "prioritize monetization in Q3, retention in Q4."
Stale propagation walks backward through →depends-on c-12. 17 cards get flagged: retention OKRs, sprint plans, hiring asks, marketing copy drafts. You see strikethrough + amber bars.
Walk through each flagged card. Some are still valid (just need re-asserting). Some need rewriting. A few get deleted. By Friday the substrate is consistent again — instead of a quarter of misaligned docs you discover at demo day.
You open a new thread to design the monetization upgrade flow. Agent reads c-12 (priority decision), c-19 (pricing strategy), c-22 (target persona) as ambient context — and starts from "given monetization is the Q3 focus and we're targeting mid-market, here are 3 upgrade flow patterns to consider."
You suggested "$30/mo enterprise" in one thread. Last week you committed to "$60 Power tier" in another. Draftlize files this as a gap_kind=conflict card with both source threads linked. You catch it before it ships.
When a card is concrete enough to ship, export to Linear (markdown copy, or paste the card URL). Mark the card built when shipped. Edit it later → status auto-flips to drift_detected.
By week 12 you've made ~80 decisions, ~200 statements, ~50 questions resolved. All addressable. All linked. All readable by your AI agent next quarter when you plan Q4.
Your free $5 covers a sprint's worth of thinking — enough to feel one decision reversal with stale propagation. That's usually when PMs realize this isn't just better notes.
Start free with $5