DISCOVER.
DEFINE.
SHIP.

Creators were stitching together 4–5 tools per week just to plan and post content — none of them spoke each other's language, and none of them planned ahead.

• Ran discovery with 20+ creators; mapped the weekly content-planning JTBD and the moments tools dropped them.
• Teardown of 8 competitors (Buffer, Later, Hootsuite, ContentStudio) — identified the natural-language-to-calendar gap nobody owned.
• Scoped MVP around one AI loop instead of feature parity; prioritized backlog using RICE with engineering.

Day-traders need to read a fast-moving market in seconds. Existing dashboards drowned them in charts and they missed trade windows watching the wrong cell.

• Defined the JTBD as "parse signal faster than the market moves" and set glance-time-to-decision as the north-star metric.
• Pushed back on stakeholder requests that added density without moving the metric; kept the surface ruthlessly small.
• Wrote the data-pipeline spec with engineering — including failure modes and degraded states so the UI never lies during an outage.

Students learning math want help on the problem in front of them right now — not a tutorial, not a search bar, not a paywall on step one.

• Reframed the JTBD as camera-first after watching 12 students fail to find the "snap" feature buried in a tab.
• Re-scoped against Apple's Human Interface Guidelines post-rejection; partnered with eng on a compliant onboarding flow.
• Defined a gamification loop tied to session-return as the growth metric — every feature had to defend itself against that bar.