1 min lesson
Make the narrative lead, not the clicking
Connect "Cut anything that doesn't move the job-to-be-done forward, even if it's a cool feature" to the decision you would make.
Step 1 of 3
Make the narrative lead, not the clicking
- Cut anything that doesn't move the job-to-be-done forward, even if it's a cool feature.
Learn more
Advanced table
Handle the skeptic honestly
Handle the skeptic honestly
Someone will push: "How is this different from Copilot?" or "Doesn't it hallucinate on a big repo?" Treat the objection as a gift. Concede what's true, then show the specific thing Cursor does better.
- Objection
- "This is just Copilot."
- Weak answer
- "No, Cursor is way better at everything."
- Credible answer
- "Copilot completes lines well. Watch what happens when the change spans four files - that's where the agent and codebase context separate."
- Objection
- "It'll hallucinate on our monorepo."
- Weak answer
- "It's very accurate, don't worry."
- Credible answer
- "It will sometimes. That's why the review step matters - here's how I'd verify a multi-file diff before accepting it."
- Objection
- "My team won't change tools."
- Weak answer
- "Everyone switches eventually."
- Credible answer
- "Adoption is the real risk. The wedge is Tab - it earns trust in a day without changing how anyone works."
| Objection | Weak answer | Credible answer |
|---|---|---|
| "This is just Copilot." | "No, Cursor is way better at everything." | "Copilot completes lines well. Watch what happens when the change spans four files - that's where the agent and codebase context separate." |
| "It'll hallucinate on our monorepo." | "It's very accurate, don't worry." | "It will sometimes. That's why the review step matters - here's how I'd verify a multi-file diff before accepting it." |
| "My team won't change tools." | "Everyone switches eventually." | "Adoption is the real risk. The wedge is Tab - it earns trust in a day without changing how anyone works." |
Honesty about limits reads as product fluency, not weakness.
Record the run and watch it back with a stopwatch. Mark every "um," every "so let me just," and every second where your hands led and your mouth lagged. Cut filler, tighten the open and confirm the whole thing lands inside the window with the objection still answered. Re-run until the narrative is the same crisp shape every time, because under panel pressure you default to your rehearsed version.
Have a backup workflow staged in a second window. Live demos stall: a model times out, a repo won't load, an extension misbehaves. If your only workflow breaks, you're improvising in front of engineers. Pick a second job-to-be-done you can run cold, keep the repo open and ready and rehearse the pivot so a stall costs you ten seconds, not the round.