If you freelance on React Native, your build time isn't where you think it is. It's not in component code — it's in the workflow loop around it: pre-kickoff scaffolding, Figma-to-real-app translation, scope-change re-quotes. This post walks through a freelance delivery workflow I've been running with an AI app builder (RapidNative) and the actual tradeoffs I've hit shipping client projects with it. Real numbers from my last 12 client projects (solo, $80/h blended, 4-week MVPs):
| Phase | Avg hours | What I was actually doing |
|---|---|---|
| Pre-kickoff scaffolding | 18h | Expo init, navigation, theme, auth screens, basic state |
| Discovery + design alignment | 14h | Figma walkthroughs, "what does this look like on a phone" debates |
| Initial screen build | 35h | 6–8 screens, basic CRUD, mock data |
| Revisions round 1 | 10h | Translating Figma feedback into code |
| Mid-project scope change | 14h | Re-quoted, sometimes absorbed |
| API wiring + polish | 18h | The actually interesting work |
| Total | ||
| ~109h |
The boring 70% is the first three rows. AI is genuinely good at those.
Before the kickoff call, I paste the client's brief into RapidNative. ~10 minutes later there's a working Expo project on a QR code. No npx create-expo-app, no theme tokens, no navigation wiring. The generated project includes Expo Router navigation, a basic design system, screens implied by the brief, and realistic mock data.
This is the step that costs me 18 hours on a traditional project. It now costs me 10 minutes plus a coffee.
The QR code matters more than the build itself. The client scans, the app opens on their phone, they tap through their own app during the call. When they want a change — "this button should be green," "remove this onboarding screen" — I use the point-and-edit mode. Click the element, type the change, watch it update on their phone in real time.
This compresses what used to be a week of feedback loops into one call. The technical reason: feedback against a real interactive app is concrete. Feedback against Figma is theoretical.
Between calls, I invite the client into the project as a collaborator. They can comment on specific screens or drop their own prompts for small changes. I wake up to a punch list with screenshots attached to the specific screen.
If they're non-technical, point-and-edit is dangerous to give them write access on — but a comment-only role is fine. The week-3 "can we add a referral flow?" used to cost me 14 hours of build + 2 hours of awkward emails. Now it's a prompt — "add a referral flow with a share button on the profile screen, a stats screen, and a redemption screen" — and an afternoon of polish. I usually absorb it. The math works because the cost is now hours, not days.
When the prototype is signed off, I export to a real React Native + Expo repo and continue in VS Code:
The export is a normal Expo project. npx expo start runs it locally. No vendor lock-in.
Well:
Not well:
The split is convenient: the "well" column is the part of freelance projects that's hardest to bill for. The "not well" column is the part you should be billing premium rates on anyway.
If a 4-week MVP becomes a 1-week sprint, do you cut your price 75%? No. You either: The speed itself is the value. Most clients pay for outcome, not hours.
The fastest test: take your next client brief, paste it into RapidNative, and see what comes back in 10 minutes. If it's useful, you saved 2 days of scaffolding. If it's not, you spent 10 minutes.
I've found it's reliably useful for the 70% of freelance work that's the same on every project. Your mileage with the other 30% — the custom business logic that's why the client hired you — is up to you.
If you've already explored this space, the export pipeline post goes deeper on what the output actually looks like.