Cycling through all six. Tap any point to stop.

Measured on six things

AirBuild is an AI app builder for native iOS apps. AI agents build your app, submit it to the App Store and handle App Review.

Security

HTTPS enforced, SPF/DMARC set, OAuth-based MCP with per-app revocation. But no HSTS, no CSP, no frame protection, and no published SOC 2, ISO, or bug bounty — 6/12 on the header checklist.

Screenshots

1 / 2

AirBuild

About this app

AirBuild (air.build) is an AI app builder for native iOS apps, made by YellowPlus, Inc. Describe your idea and a team of AI agents for product, design, engineering and growth builds a working iOS app. AirBuild then submits it to the App Store, responds to App Review, fixes rejection issues and resubmits.

Key features

1AI agent team (Product, Design, Engineering, Growth) that builds a native iOS app from a brief.
2Automated App Store submission, App Review response, rejection fixes and resubmission.
3MCP server for driving AirBuild from Claude, ChatGPT, Cursor or Codex via OAuth.

Target customer profile

Drawn from the product itself, not from a survey.

Demographic

Founders and designers who want to turn an app idea into a working native iOS app

Demographic

Developers who want AI agents to take over App Store submission and App Review back-and-forth

Demographic

Small businesses that want their own iPhone app without hiring a full team

Demographic

Teams moving an existing GitHub project or a web prototype to a native iOS app

Scout Review

Written by AI from measured evidence, scored out of 100.

AirBuild is an AI app builder with a genuinely differentiated pitch: not just generate a native iOS app, but submit it to the App Store, respond to App Review, fix rejections and resubmit automatically. The product ships as an iOS app with a distinctive agent-avatar UI and a clear four-agent team model (Product, Design, Engineering, Growth). The marketing site is the weak link — inconsistent CTAs, dead space, a truncated card — while the MCP docs page shows the team can write real documentation when it wants to. Speed is excellent on desktop and good on mobile; security headers are thin; accessibility has real gaps. The idea is the strongest part.

Measure by measure

Usability

Clear four-step Submit/Respond/Fix/Resubmit model and a genuinely detailed MCP docs page. But the homepage offers 'Download' and 'Join the waitlist' with no guidance on which, and the product mock-up text is unreadable.

Design

Judging the app UI from store screenshots: the agent-avatar system is distinctive and the list hierarchy is clean. The marketing site is weaker — inconsistent CTA colors, dead space, a truncated card.

Speed

Desktop Lighthouse 100/100, mobile 88/100 with TBT 19ms and FCP 1.1s. The 3.4s mobile LCP is the one soft spot — above the 2.5s 'good' line, likely a heavy hero image.

Security

HTTPS enforced, SPF/DMARC set, OAuth-based MCP with per-app revocation. But no HSTS, no CSP, no frame protection, and no published SOC 2, ISO, or bug bounty — 6/12 on the header checklist.

Accessibility

Lighthouse 81/100 with real failures: unnamed buttons, low contrast, small touch targets, ARIA role mismatches. Localized in 13 languages, which is a genuine strength, but no accessibility statement anywhere.

Growth

Real, specific pain — App Store submission and review back-and-forth — with a wedge Lovable and Bolt.new structurally don't cover: native iOS plus automated App Review. Early (domain 0.4y, 3 ratings) but the idea has room.

Improvement areas

Inconsistent CTA styling and truncated cards on the homepage undermine a strong product pitch.
Missing HSTS, CSP and frame-protection headers; no published SOC 2 or bug bounty.
Accessibility gaps: unnamed buttons, low contrast, small touch targets, ARIA role errors.

Design & Usability

The app itself, judged from store screenshots, has a coherent identity: agent avatars with consistent iconography and a readable name/handle/status list. The marketing site doesn't match that discipline — three CTA colors for one action, dead bands between sections, a truncated rankings card, and a product mock-up whose text is too small to read. Usability is stronger: the four-step Submit/Respond/Fix/Resubmit row gives an instant mental model, and the MCP docs page is unusually thorough with per-client setup for Claude, ChatGPT, Cursor and Codex. The homepage undercuts itself by offering 'Download' and 'Join the waitlist' side by side with no guidance on which path applies to whom.

Speed & Security

Speed is a tale of two devices: desktop Lighthouse 100/100, mobile 88/100 with TBT at 19ms and FCP at 1.1s — genuinely snappy. The 3.4s mobile LCP is the outlier, above the 2.5s 'good' threshold and the main thing holding mobile back. Security is the weakest measured dimension: HTTPS is enforced and SPF/DMARC are configured, but there's no HSTS, no CSP, no frame protection and no Permissions-Policy — 6/12 on the header checklist. The MCP server's OAuth flow with per-app revocation under 'Connected apps' is a thoughtful touch. No SOC 2, ISO or bug bounty is published, which is normal for this stage but leaves the compliance story thin.

Accessibility & Growth

Accessibility lands at 81/100 on Lighthouse with concrete failures: buttons without accessible names, insufficient color contrast, touch targets too small, and ARIA roles missing required children. The 13-language localization is a real strength, but there's no accessibility statement and the automated score only covers part of WCAG. Growth is the most interesting dimension: AirBuild names a specific, painful problem — App Store submission and the App Review back-and-forth — and automates it end to end, which is a wedge that Lovable and Bolt.new structurally don't cover since they generate web apps, not native iOS with store submission. The category has headroom as AI app builders proliferate. Traction is early (domain 0.4 years old, 5/5 from 3 ratings), which is context, not a verdict on the idea.

Conclusion

If AirBuild delivers on the App Review automation it's promising, it solves a problem every iOS indie dev knows intimately and nobody enjoys. The pricing is sensible — Free with BYOK, $20 Pro, $200 Max, $2,000 Enterprise — and the MCP server integration is a smart bet on where AI-assisted dev is heading. The gaps are fixable: tighten the landing page, add security headers, name your buttons for screen readers. The core bet is worth watching.

Suggestions for improvement

Unify the CTA system: pick one color and one label for the primary action across nav, hero and product window.
Fix the truncated 'Track rankings' card and reduce the dead vertical space between homepage sections.
Add HSTS, CSP and frame protection headers — you're at 6/12 on the security checklist.
Give every button an accessible name and raise contrast/touch-target sizes to clear the failing Lighthouse audits.

How we score

Alternatives comparison

Named competitors, point by point. Nobody paid to appear here or to be left out.

	AirBuild	Lovable	Bolt.new
Primary output	Native iOS app with App Store submission	Web app / full-stack web product	Web app in browser sandbox
App Store submission automation	Yes — submits, responds to review, fixes and resubmits	No	No
AI agent team model	Product, Design, Engineering, Growth agents sharing one brief	Single AI assistant	Single AI assistant
MCP server integration	Yes — OAuth remote MCP for Claude, ChatGPT, Cursor, Codex	Not advertised	Not advertised
Free tier	Free with BYOK, 1 hr cloud VM/month	Free tier available	Free tier available
Entry paid price	$20/mo Pro	Subscription tiers	Subscription tiers