🎯 For CreateOS specifically: The one-sentence version of this playbook is in CreateOS - Overview. Your positioning is "the only agent sandbox where the agent can earn." Your 30-day BD priority is the OpenAI Agents SDK inclusion (contact: Karan Sharma). Your launch is Agent Economy Week in month 2-3. Everything below is the general framework; the CreateOS-specific execution lives in CreateOS - 30-60-90 Plan.
The thesis, in one sentence
The UX shells of AI app-builders have converged; the winners now differentiate on audience focus, monetization shape, and one signature pattern executed better than anyone.
Top 5 moves, ranked by impact
1. Install a pre-signup prompt box on the landing page
Observed: Both Replit and Anything.com. Why it works: converts "visitor reading copy" into "user with an idea waiting for signup." 2-3× activation multiplier vs. a plain free-trial CTA.
2. Meter usage with credits + show consumption persistently
Observed: Replit's left-sidebar usage widget (Free Apps / Agent credits / Cloud credits). Why it works: transparency builds trust (no surprise bills → no churn) and creates upgrade urgency (watching a bar fill is its own CTA).
3. Pick one audience ring and over-serve it
Payments + App Store Marketing + Monetization for makers (Anything's stack). SBOM + SOC 2 + admin hierarchy for enterprise (Replit Core). Code editor + palette + version control for devs. Don't straddle all three unless you're at Replit scale.
4. Productize autonomy as a distinct purchase moment
Bundle: Replit Core (Plan + Connectors + Autonomous + Simulators + Background). Unbundle: Anything Max. Both work. What doesn't work: giving autonomy away free — users pay for it if you name and scope it.
5. Ship one viral free-tier loop
"Made with X" watermark + credit reward is Replit's. Anything.com has nothing equivalent despite a maker ICP that would love it. Every free-tier app should be a distribution channel.
The 2×2: do / avoid by objective
| ✅ Do | ❌ Avoid | |
|---|---|---|
| Acquisition | Pre-signup prompt · competitive-named imports · real customer logos · distinctive brand imagery · example-prompt chips | Utility-only landing · no social proof · hidden features · generic "try it free" CTA |
| Retention | Transparent usage widget · named design presets · visible agent reasoning · in-flight stop button · rich version history · command palette | Hidden consumption · opaque agent work · no stop during generation · blank "your recent projects" grid |
| Revenue | Credit metering · gate-at-moment-of-desire paywall · bundle OR unbundle autonomy (not both) · branded watermark loop · monetization-stack bundle (for makers) | Flat pricing · paywalls in settings pages · selling autonomy for free · no consumption visibility · no enterprise-procurement artifacts |
| Strategy | One audience · one autonomy axis · one signature pattern executed better than peers · one viral loop | All-three-rings positioning · autonomy-breadth + depth + planning (pick 1-2) · matching peers feature-for-feature |
Where to read next
- 01 - Acquisition — landing page, imports, social proof, first-run flow
- 02 - Retention — usage widget, agent reasoning, design systems, version history
- 03 - Revenue — credit metering, tier shape, paywall placement, viral loops
- 04 - UX Recommendations — patterns to adopt + anti-patterns to avoid
- 05 - Strategic Bets — the hard tradeoffs
How to measure (top-level)
| Metric | Target behavior | Proxy |
|---|---|---|
| Visitor → active user | 25-40% of visitors land on the prompt box and type something | Landing prompt submit rate |
| Signup → 7-day retention | 50%+ return within 7 days | Users with ≥ 2 sessions in week 1 |
| Free → paid conversion | 3-8% of active free users upgrade within 30 days | Paid conversion at day 30 |
| NRR on paid base | > 110% annualized | Paid cohort revenue month over month |