AI ROI Calculator
Estimate AI tool subscription and API token costs against hours saved, hourly labor value, and setup payback months. Free browser-based calculator, no sign-up.
Open toolAI App Playbook: Niche Scorecard, Checklist & Cost Calculator
Score your consumer AI app idea against the 7-criterion Niche Scorecard (25/35 pass threshold), calculate your $2.70/download subscription funnel and max CAC ceiling, compare the Cloudflare + DeepSeek Lean Stack against Vercel + Supabase + Clerk, and export a ready-to-run AGENTS.app.md launch brief.
7-Criterion Niche Scorecard$2.70/Download Funnel MathLean vs. Popular Stack Costs15-Phase AGENTS.md Checklist
Enter your details
Enter the inputs, then select Evaluate Playbook & Unit Economics.
Your result
Enter your details
Result will appear here
What Is the AI App Playbook (and How This Calculator Models It)?
Published on October 26, 2026 by Blake Anderson (@blakeandersonw, founder behind consumer AI apps Cal AI, Umax, PlugAI, and EcoGPT), the open-source AI App Playbook (github.com/blakeandersonw/ai-app-playbook) is an opinionated operating brief for solo founders and AI coding agents (Claude Code or OpenCode). Instead of treating an AI app as a complex enterprise SaaS, it pairs a strict 7-criterion niche filter with a low-overhead Cloudflare + DeepSeek + Superwall stack so total infrastructure and LLM costs stay near 2.5% of gross revenue.
RevPerDownload = TrialStartRate * TrialToPaidRate * AnnualPrice ($0.30 * $0.30 * $30 = $2.70); MonthlyCalls = (PaidSubs * 60) + (UnpaidTrials * 15) + (NonTrialInstalls * 2); CallCost = (1500 * InRate + 400 * OutRate) / 1e6 * 1.055.
If you are evaluating an internal company AI copilot or workflow automation rather than a consumer subscription app, use our AI ROI Calculator (/tools/ai-roi-calculator) to compare monthly seat/API costs against engineering or operations hours saved.
AI App Playbook Reference Table: 15 Phases, Lean Stack vs. Popular Stack & Token Costs
Benchmark numbers from the 2026 AI App Playbook (github.com/blakeandersonw/ai-app-playbook) at 100,000 MAU and 1,500 input + 400 output tokens per AI call.
| Playbook Dimension / Layer | Playbook Lean Standard (2026) | Popular Default Alternative | Cost / Funnel Benchmark & Rule |
|---|---|---|---|
| Phase 0: Niche Scorecard | 7 Criteria (1 to 5 scale, Total >= 25/35) | Building before demand validation | Zero criteria may equal 1; validate with 10-15 short videos + >10% waitlist CVR. |
| Subscription Funnel Math | 30% Trial Start x 30% Paid @ $30/yr | Freemium without paywall trial | 9.0% net download-to-paid = $2.70 revenue per download (sets max profitable CPI). |
| Hosting, Edge API & Jobs | 1 Cloudflare Worker ($0.21 / 1k MAU) | Vercel Pro + Next.js ($3.48 / 1k MAU) | ~17x cheaper on Workers; avoids 10s open streaming connection duration billing. |
| SQL Database & Object Storage | Cloudflare D1 ($0) + R2 ($0.15 / 1k MAU) | Supabase Pro + Storage ($1.62 / 1k MAU) | ~11x cheaper; plain parameterized SQL in D1 (no ORM) and zero R2 egress fees. |
| Authentication Layer | Better Auth in Worker ($0.00 / 1k MAU) | Clerk ($10.25 / 1k MAU at 100k scale) | Anonymous session on first launch; link Apple/Google only after paywall conversion. |
| AI Model Routing (60 calls/mo) | DeepSeek V4.1 Flash ($0.49 / 1k calls) | OpenAI gpt-6.1-sol ($420 / 1k MAU) | $29.43/1k MAU on DeepSeek Flash vs $420/1k MAU (~14x cheaper; 70x vs gpt-6-astra). |
| Analytics, Errors & Paywall | Tinybird ($1.45/1k) + Superwall (1% rev) | PostHog + Sentry ($4.97/1k MAU) | Total Lean Stack + AI stays locked at ~2.5% of gross revenue from 1k to 1M downloads. |
How to Use the AI App Playbook Calculator in 3 Steps
Run your consumer AI app concept through the exact Phase 0 scorecard and Chapter 12 cost formulas before writing application code.
Score your app idea across the 7 Niche Criteria (1 to 5)
Rate Pain/Desire, Usage Frequency, 3-Second Video Demoability, AI as the Core Engine, $30/yr Willingness to Pay, Reachable Community (TikTok/Reels/Reddit), and Low Regulatory/Liability Risk from 1 to 5. The calculator checks both the >= 25/35 pass threshold and the single-1 veto rule.
Enter your download scale, paywall rates, and LLM routing tier
Set your monthly downloads (1,000 to 1,000,000+), trial start percentage (30% benchmark), trial-to-paid conversion (30% benchmark), and annual price ($30 default), then pick your OpenRouter model tier (DeepSeek V4.1 Flash, DeepSeek V4 Pro, Gemini 3.8 Flash Vision, or Claude Sonnet 5.5).
Audit Lean vs. Popular stack burn and export your AGENTS.app.md brief
Review your gross monthly revenue, maximum profitable CAC per download, Lean Stack vs. Popular Stack monthly savings, after-store-fee net contribution (85% Apple/Google Small Business tier), and copy the generated AGENTS.app.md project brief.
Worked Examples from the 2026 AI App Playbook
See how the Playbook math behaves across a 10,000-download vision scanner, a Superwall paywall lift experiment, and a Phase 0 scorecard veto.
Consumer Vision Scanner at 10,000 Downloads/Month (Cal AI / Yuka Archetype)
Scoring [5, 5, 5, 5, 4, 5, 4] clears Phase 0 at 33/35. At 10,000 monthly downloads, 30% start a trial (3,000 trials) and 30% convert to paid (900 subscribers at $30/yr), generating $27,000/month ($2.70/download). Lean Stack costs total $690/month ($270 Superwall 1% + $20 Cloudflare + $50 Tinybird + $350 AI processing), leaving $26,310/month (97.4% pre-store margin, or $22,260/month after the 15% App Store Small Business fee).
Phase 14 Funnel Experiment: Lifting Trial Starts from 30% to 40%
Keeping downloads fixed at 10,000/month and trial-to-paid at 30%, moving the core AI camera demo into onboarding so trial starts rise from 30% to 40% increases paid subscribers from 900 to 1,200/month. Monthly gross revenue jumps by +33.3% from $27,000 to $36,000 ($3.60/download), while incremental Superwall and AI token costs add less than $165/month.
Phase 0 Veto Check: Why a 26/35 Medical or Legal Wrapper Gets Rejected
An AI symptom or legal contract adviser scoring [5, 4, 4, 5, 4, 3, 1] reaches 26/35 in total points, above the 25-point cutoff, but scores 1 on Criterion 7 (Low Regulatory/Liability Risk). The Playbook enforces a hard veto on any criterion equal to 1 because App Store Guideline 1.4.1 rejections and clinical liability stall distribution.
How to Interpret Your Playbook Numbers & Stack Architecture
The central insight of Chapter 12 is that infrastructure and LLM inference should never be the reason an indie consumer app fails.
First, treat revenue per download ($2.70 at the 30% x 30% x $30/yr benchmark) as your hard ceiling for Customer Acquisition Cost (CAC per install). If short-form creator UGC on SideShift or Trybe costs $0.80 to $1.40 per install, you are profitable in month one; if paid ads cost $4.00 per install, you must either raise trial starts above 40% or increase annual pricing before scaling spend. You can model multi-year retention cohorts in our Customer Lifetime Value Calculator (/tools/customer-lifetime-value-calculator) and Churn Rate Calculator (/tools/churnratecalculator).
Second, route 90%+ of routine text calls through DeepSeek V4.1 Flash ($0.15/$0.60 per 1M tokens, or $0.49 per 1,000 calls of 1,500 input + 400 output tokens including OpenRouter's 5.5% fee). At 60 calls per subscriber per month, AI inference costs just $0.0294 per paying user against $2.50/month of subscription revenue. Reserve Gemini 3.8 Flash ($3.05/1k calls) for camera/vision inputs and Claude Sonnet 5.5 ($7.38/1k calls) only where blind evals prove users notice the difference.
15-Phase AGENTS.md Checklist, Anti-Bloat Rules & Related Tools
When pointing Claude Code or OpenCode at AGENTS.md, enforce these four architectural guardrails from Phases 1 through 13.
1. What the Playbook Explicitly Avoids ('Do Not Use' List): Skip Vercel, Netlify, Next.js, ORMs, Lovable, Bolt, v0, Replit hosting, Bubble, FlutterFlow, Firebase, Supabase, Clerk, Auth0, Sentry, and PostHog. A single Cloudflare Worker with plain parameterized SQL on D1, zero-egress R2 storage, Better Auth running inside the Worker, and Tinybird for both analytics and error logs eliminates per-seat and per-MAU pricing cliffs.
2. Mandatory Pre-Launch Security & Spend Caps (Phase 12 & 13): Always start users with an anonymous Better Auth session so they experience the core AI action in under 10 seconds before creating an account, then enforce hard daily per-user and global USD spend caps in your Worker middleware alongside Tinybird ai_call.cost_usd logging so a bot or prompt loop cannot drain your OpenRouter balance.
Customer Lifetime Value Calculator
Calculate Customer Lifetime Value (CLV), gross lifetime revenue, LTV:CAC ratio, and CAC payback months from AOV, frequency, margin, and lifespan.
Open tool
Conversion Rate Calculator
Find conversion rate from unique eligible participants and goal-completing participants.
Open toolMore free tools to try
Choose another tool for your next calculation, conversion, or text task.
ASCII to Text Converter
Validate whitespace- or comma-separated integer codes from 0 to 127 and map them to ASCII characters.
Open tool
Base32 Encoder Decoder
Encode UTF-8 text or hex bytes as RFC 4648 Base32, or decode canonical Base32 into hex, text and a byte file.
Open tool
JSON to C# Class Generator
Generate nested C# model definitions from a JSON example. Review the inferred fields, then copy or download the source.
Open tool
JSON to Go Struct Generator
Generate nested Go model definitions from a JSON example. Review the inferred fields, then copy or download the source.
Open toolFrequently Asked Questions About the AI App Playbook
Answers about using AI App Playbook: Niche Scorecard, Checklist & Cost Calculator and understanding its results.
What is the AI App Playbook on GitHub?
The AI App Playbook (github.com/blakeandersonw/ai-app-playbook) is an open-source 14-chapter guide and agentic template repository released in October 2026 by Blake Anderson (founder of Cal AI, Umax, PlugAI, and EcoGPT). It documents the exact niche scorecard, Cloudflare + DeepSeek + Superwall tech stack, $2.70/download unit economics model, and short-form UGC distribution workflow used to scale consumer AI apps to $100k+ MRR.
How does the 7-Criterion Niche Scorecard work in Phase 0?
You rate an app idea from 1 to 5 across seven criteria: (1) Pain or Desire, (2) Daily/Weekly Frequency, (3) 3-Second Video Demoability, (4) AI as the Core Engine, (5) Willingness to Pay $30/year, (6) Reachable Community on TikTok/Reels/Reddit, and (7) Low Regulatory/Liability Risk. You only proceed to build if the total score is 25 or higher out of 35 AND no single criterion scores a 1.
How is the $2.70 revenue per download benchmark calculated?
The Playbook models a standard consumer subscription funnel where 30% of app downloads start a free trial on the paywall, 30% of those trials convert to paid (30% x 30% = 9.0% net conversion), and subscribers pay $30/year. Multiplying 9.0% by $30 equals $2.70 in gross revenue per download ($2,700 at 1,000 downloads; $27,000 at 10,000 downloads; $270,000 at 100,000 downloads).
Why does the Playbook choose Cloudflare Workers, D1, Better Auth, and DeepSeek over Vercel, Supabase, and Clerk?
At 100,000 monthly active users (500 API requests, 10 MB storage, and 60 streamed AI calls per user), Vercel Pro charges for the ~10-second open streaming duration ($3.48/1k users vs $0.21 on Cloudflare Workers), Clerk charges $0.02 per MAU above 50k ($10.25/1k users vs $0 for Better Auth inside a Worker), and legacy frontier LLMs cost $420/1k users vs $29.43/1k users on DeepSeek V4.1 Flash. Combined, the Lean Stack costs ~$32.28 per 1,000 users/month compared to ~$442.47 on the popular stack.
How do I use this Playbook with Claude Code or OpenCode?
Clone github.com/blakeandersonw/ai-app-playbook into an empty folder, fill out your AGENTS.app.md brief (or export the brief generated by this calculator), and instruct Claude Code or OpenCode to read AGENTS.md. The agent steps through Phases 0 to 14 sequentially, verifying each phase's 'Done when' checklist before moving to the next.
When should I use the AI App Playbook Calculator vs. the AI ROI Calculator?
Use this AI App Playbook Calculator when you are building and monetizing a consumer or prosumer AI app (modeling download-to-trial funnels, $30/yr subscriptions, OpenRouter token costs, and Cloudflare stack burn). Use our AI ROI Calculator (/tools/ai-roi-calculator) when your company is buying AI seats or API credits internally and wants to measure payback months from employee hours saved.
Try your own values
Choose another tool for your next calculation, conversion, or text task.