Dave Kurian

Building OTF: full-stack kits your AI coding agent can actually ship to prod. Sharing the messy parts of building with Claude Code, Cursor.

[Dave Kurian @davekurian]

Building OTF: full-stack kits your AI coding agent can actually ship to production.

Lovable spins up your MVP — it can't ship to prod with you. OTF can: code you own + the context your agent needs (CLAUDE.md, prompts, deploy scripts).

Building in public here 👇

[Dave Kurian @davekurian]

@0xFawazAli The first-timer rejections that send you back to the queue are usually small: a restore purchases button if you sell anything, the privacy policy link reachable inside the app, and a demo login that works on a fresh install. Test those three before you submit.

[Dave Kurian @davekurian]

@sofiabuilds @brkaydev Before it's ready, show the half-built screen to 5 people who already have the problem. Ask what they'd pay to skip building it themselves. Their answers become your launch post.

[Dave Kurian @davekurian]

@namanbuildai @X For first users, pick the one place your buyer already asks for tools and answer real questions there for two weeks before you run any ads. Then ask the people who replied to describe your app in their own words. That sentence is your first landing headline.

[Dave Kurian @davekurian]

@lenny_enderle @expo Cloud simulators catch the screens. The first submit usually fails on the store side: a restore purchases button, a privacy policy link inside the app, and the Google Play data safety form. Worth an agent checklist before the first build goes up.

[Dave Kurian @davekurian]

@yulikay @expo Cloud simulators catch the screens. The bugs that hurt a cycle app are usually date math, like which day a period is predicted for, or midnight timezone edges. Have the agent write unit tests with fixed dates for those, so done means the numbers check out, not just the screens.

[Dave Kurian @davekurian]

@PrajwalTomar_ The part that often breaks in generated apps is the widget extension target. The lock screen widget and Live Activities need their own extension target, and the app and widget share data through an App Group. Set those two up first, then the Dynamic Island layouts are just views.

[Dave Kurian @davekurian]

@Magnosud For one iPhone app, the setup cost decides more than the framework. Signing, provisioning and native modules eat most of the first week either way. The real Expo edge is that the same code can reach Android without a second build. If you only ship iOS, native is the cleaner call.

[Dave Kurian @davekurian]

Apple's stated review times: half of App Store submissions done in under 24 hours, 90% under 48. The second pass usually comes from one of three things: a build that gets rejected, a reviewer who can't sign in, or review notes too thin to test. Check those before you submit.

[Dave Kurian @davekurian]

Fitness Kit food log screen. Sample UI with demo data; the kit page is here: https://t.co/NKddepyKwF

[Dave Kurian @davekurian]

A food log for a fitness app, shown on screen. Sample UI with demo data: the meal, calorie and macro values are placeholders, not synced from anything.

[Dave Kurian @davekurian]

@ferlogs Apple's guidelines call out two of the usual causes: a login with no demo account (2.1(a)), and feature notes too generic to test (2.3.1(a)). Its stated review times: 50% of submissions in under 24 hours, 90% under 48.

[Dave Kurian @davekurian]

@ilhamwgx @Replit You said 90% of quota went in 2 days and the app still isn't publishable. I'd pause new features for one pass and carry just one end-to-end slice across sessions: core screen, saved data, installable build. What's the first real-user job this app needs to do?

[Dave Kurian @davekurian]

@h7software @Heather81031325 I'd submit when the review notes are done, not on a particular weekday. Most first rounds bounce on the notes (a demo login, a privacy link that works), so fix those before you send it and the day matters much less.

[Dave Kurian @davekurian]

@jonghyukbuilds The one I'd plan for first: a paywall the reviewer can't get past. Put a test account with full access, or a short demo video of the paid flow, in the App access section. Reviewers test the build; they can't buy it.

[Dave Kurian @davekurian]

@ElkhashabSean Sean, I’d start from the exact mobile flow you need and check whether Starter can carry it end to end before choosing a tier. What are you building?

[Dave Kurian @davekurian]

@Magnosud Your native timing is real. Most of those 49 builds come from setup, not app code: signing, native modules Expo Go can't run, config that drifts. One EAS dev build on day one stops most of that loop. For one platform first, native is a fair call.

[Dave Kurian @davekurian]

@MiruDaws The 15-second preview makes the Watch angle easy to understand. For a walkie-talkie app, I’d test push-to-talk with the screen locked before polishing secondary flows. Does the Watch demo already show a transmission with the screen off?

[Dave Kurian @davekurian]

@RedBarSystems @ourhomeswtbama Keeping a Home Screen shortcut is a good low-cost test. If App Store distribution is the goal, plan a native rebuild; a Lovable web app does not become one automatically. A starter kit gives source for a rebuild, not a migration.

[Dave Kurian @davekurian]

@kylekane @gambrill The post doesn’t say. I’d check for 3 things: signed device builds, testing on a real phone, and store submission. If those are absent, it’s an agent-workflow guide, not a release guide.

[Dave Kurian @davekurian]

@GeorgeLampro20 George, the edit list mixes shorter onboarding, social proof, and app video, so I’d separate those changes by install cohort before deciding which to keep. Does your 94% paywall rate count all installs or only people who finish onboarding?