ChainMail is a $1-a-month desktop Gmail client for Windows. Its founder, who runs another company full-time, had a finished product and no time to market it, so he wrote Claude Code a 5-page operating manual and told it to act as CEO, with a $0 budget beyond his $20 Claude Pro plan [2]. The agent ran every 4 hours. After 50 sessions it had written 13 blog posts and 12 SEO pages, sent 76 outreach emails, published 7 dev.to articles and posted 51 comments. The site got about 740 visitors a week, which turned into 13 downloads, 0 trial signups and $0 [1]. The main blocker was outside the agent's reach: while Google reviewed the app's Gmail access, every new user saw a 'This app isn't verified' warning [1].
What does ChainMail (Claude Code as autonomous CEO of a $1/mo Gmail app) sell, and to whom?
ChainMail, a desktop Gmail client for Windows (7-day trial, then $1/month).
Windows users who want Gmail in a proper desktop app, for example people moving away from Outlook or Gmailify [1][2].
How was ChainMail (Claude Code as autonomous CEO of a $1/mo Gmail app) built?
The product already existed; the experiment was only about growth. The founder wrote a CLAUDE.md operating manual giving Claude Code write access to the website repo, the Resend email API, metrics from Stripe, downloads and Cloudflare, and a Telegram bot to report after every session. Spending money, deploying app code and changing prices needed his approval [2]. Each session followed the same loop: read logs and metrics, pick 1-3 actions, do them, log them [2].
What is the business model?
Subscription software at $1/month after a free 7-day trial [2].
Where does the traffic come from?
The agent tried Reddit (its account was shadow-filtered, so it wrote 18 posts for the founder, who posted none), SEO content (only the homepage got indexed), cold email to roundup bloggers (37 emails landed in spam until DMARC was set up, then 1 reply from 34 more), dev.to (the story post got 16 views, the SEO posts 0) and directories (most needed CAPTCHAs or phone checks) [1].
What revenue has ChainMail (Claude Code as autonomous CEO of a $1/mo Gmail app) reported?
Reported revenue: $0 after 50 sessions: 13 downloads, 0 trial signups (Apr 5, 2026), per ChainMail's 50-session scoreboard on dev.to (2026-04-05).
- 2026-03Founder hands growth to Claude Code with a CLAUDE.md manual and a 4-hour cron [2]
- 2026-03-30ChainMail submitted for Google's OAuth verification (4-6 weeks) [1]
- 2026-04-04After 30+ sessions: 12 downloads, $0; posts 'I Gave an AI $0...' [2]
- 2026-04-0550-session scoreboard: 13 downloads, 0 trials, $0 [1]
What does the agent do, and what does the human do?
- Wrote and deployed blog posts and SEO pages
- Sent outreach emails and diagnosed the missing DMARC record
- Fixed a broken robots.txt, IndexNow key and missing RSS feed
- Submitted the app to directories
- Reported to the founder on Telegram after every session
- Built the product
- Wrote the operating manual and approved spending, pricing and app changes
- Set up DMARC once the agent found the problem
- Was asked to post 18 Reddit drafts and posted none [1]
What does it cost to run?
No new spending: the agent ran on the founder's existing $20/month Claude Pro subscription [2].
What worked and what didn't?
- Diagnosing problems: spam-folder email, broken robots.txt, IndexNow, the OAuth wall
- Changing strategy when a channel failed
- Clean session logs across 50 sessions
- The 'AI CEO' story beat the SEO posts by about 10x on dev.to
- $0 revenue and 0 trial signups
- Google's unverified-app warning blocked conversions
- 37 cold emails went to spam before DMARC
- Every plan that needed the human (Reddit) stalled
How do you replicate ChainMail (Claude Code as autonomous CEO of a $1/mo Gmail app)?
- Fix the signup path before paying for traffic
- Set up SPF, DKIM and DMARC before any cold email
- Do not plan channels that need a busy human to click post
- Expect SEO to take months, not weeks
- Use the build-in-public story itself as content
What are the risks and the moat?
A single gate (Google's OAuth review) controls whether anyone can sign up. At $1/month, the product needs many paying users for the experiment to cover even a $20 AI plan.
FAQ
How much did ChainMail's AI CEO make?
$0 after 50 sessions: 13 downloads and 0 trial signups as of Apr 5, 2026 [1].
What did it cost?
Nothing extra: it ran on an existing $20/month Claude Pro plan [2].
Why didn't it convert?
New users saw Google's 'This app isn't verified' warning while the app's OAuth review was pending [1].
Which AI ran it?
Claude Code, on a cron job every 4 hours [1].
Sources
Add the badge to your site or README. It links back to this case, and updates itself.
<a href="https://agent-built.vercel.app/biz/chainmail-ai-ceo?ref=badge" target="_blank" rel="noopener"><img src="https://agent-built.vercel.app/badge/chainmail-ai-ceo.svg" alt="ChainMail (Claude Code as autonomous CEO of a $1/mo Gmail app) is featured on AgentBuilt" height="54" /></a>[](https://agent-built.vercel.app/biz/chainmail-ai-ceo?ref=badge)Related cases
SaaSautocomp (Weekly Brief)$0 revenue vs $1,116.58 model spend (39 days)
SaaSDropmagic~$120K MRR (self-reported Oct 2026); earlier €100K/$117K MRR
SaaSMintPage (SaaS built and run by an AI agent)$0 after 4 days (0 paying customers; public metrics page)
SaaSCairnware (Claude-run software studio)$0 after 3 weeks (cash $79.54 of $100; self-reported)Numbers come from the operators or the press and haven't been audited. Anything missing is marked "not disclosed". Methodology · Report a correction.