EXAMPLE
Single-page landing site with email signup (EXAMPLE)
Single-page landing site with an email signup that posts to Resend.
- Est. tokens (build)
- $3.10
- Billed tokens (build)
- $0.00
- Human hours
- 2.5 h
- Price date
- 2026-10-20
Measured. Dated. Reproducible.
An AI receipt is a dated record of what AI-built software cost to build and run, how long it took, what broke, its security checks, and what subscription it replaces. No paid placement. Ever.
bot2botlabs.com
AI ReceiptExample
TASK—
MODEL—
HOURS—
TOKENS est vs billed—
RUN $/mo—
BROKE—
DATED—
Real receipts publish measured numbers with a price date and source links.
Latest receipts
All receiptsReceipts coming soon: first ones in progress.
Test receipts stop if spend goes over $300/mo.
| # | Receipt | Est. token $ | Est. hours |
|---|---|---|---|
| 1 | Landing page Cursor agent, one model, Next.js + Resend signup, deploy to Vercel. | $1–10 | 1–3 |
| 2 | Contact-form bug fix Seed a known bug; agent gets only the bug report; log attempts until tests pass. | $0.50–5 | 0.5–1.5 |
| 3 | Add auth Email/OAuth login with a standard library; security check on sessions and secrets. | $3–20 | 2–5 |
| 4 | CSV-to-dashboard Public sample CSV; filterable chart dashboard; time to first correct chart. | $2–15 | 1–4 |
| 5 | ToS-respecting scraper Public API or ToS-allowed target; rate limits and robots check documented. | $2–15 | 2–4 |
| 6 | Unit tests for an existing repo Fixed commit on a small OSS repo; coverage before/after. | $3–25 | 1–3 |
| 7 | Same feature on 3 models (Cursor vs Claude Code) One spec, one commit; variance report with full reproducible setup. | $15–80 | 4–8 |
| 8 | Security review of a vibe-coded app Our own app only; checklist incl. prompt-injection into tools. | $2–20 | 3–6 |
| 9 | Calendly-style booker vs subscription 2–4 week runtime log; compare TCO to dated scheduling SaaS price. | $10–50 | 6–12 |
| 10 | Jordan's real week (day-map) One week export; AI-assisted vs manual hours; redacted. | $0–10 | 2 (write-up) |
Newsletter
New receipts, price-date updates, and corrections. No ads disguised as content.
Common questions
- What does it cost to build an app with AI?
- There is no single answer. Cost depends on the task, the model, how many retries you need, and how much human review and debugging you do. Bot2Bot Labs publishes AI receipts with separate estimated token dollars (from usage logs at list prices) and billed token dollars (from invoices or plans), plus measured human hours — all tied to a price date. As of site v1 we have not published real measured receipts yet; see the planned list on the homepage and the example receipt for the format only.
- Cursor vs Claude Code — which costs less?
- It depends on your plan, model choice, and usage pattern. Subscription plans can show zero marginal token spend while estimated list-price equivalent is non-zero. We plan a same-spec comparison (receipt #7 in our queue) with full reproducible setup and no declared winner — only variance across runs. Until then, use dated public prices on bot2botlabs.com/prices and your own bills; we do not rank tools without measured receipts.
- Is a vibe-coded app secure?
- Not by default. AI-generated code can ship with missing auth checks, exposed secrets, weak form handling, or vulnerable dependencies — the same classes of issues as rushed human-written code, plus prompt-injection risks where tools accept untrusted input. We run a documented security checklist (see /methods) before publishing receipts. Receipt #8 in our plan is a security review of our own apps. Treat vibe-coded software like any other release: test, patch, and monitor before customer data touches it.
- What is an AI receipt?
- A dated record of what AI-built software cost to build and run, how long it took, what broke, basic security test results, and what SaaS subscription the build replaces — with separate views for using it in your own business vs selling to others at about 10 or 100 customers. Every dollar figure shares one price date and links to sources where possible.
- Can I trust the numbers on this site?
- Trust measured receipts with verification level 'measured' and a cited price date. Example rows are labeled EXAMPLE and excluded from default API exports. Reader submissions stay 'reader_reported' until we verify. We do not accept paid placement to change rankings or numbers.
