A solo developer shares two-week results from selling 12 small paid tools (1–25 cents/call) to AI agents via the x402 payment standard, reporting ~7 cents in revenue from one returning buyer and lessons about bot-filtered traffic and agent-focused tool design.
Side project update, partly to think out loud and partly to hear from anyone else trying this. **The idea** AI agents are starting to pay for things on their own using x402, a standard where a web API answers "402 Payment Required" with a price, the agent pays a few cents in USDC on Base, and gets the answer. No account, no API key, no human in the loop. I wanted to see if a solo dev can make money selling small, useful tools to those agents. **What I've built so far** 12 paid tools, priced between $0.01 and $0.25 per call. All facts from primary sources, no opinions: - package_check: before an agent installs an npm/PyPI package, check for known malware, critical vulns, typosquats, removed versions and brand-new packages - why_is_it_moving: for a stock ticker, the SEC filings from the last 48h decoded into plain English (8-K items, press-release headline, offerings, insider buys and sells) plus whether it was halted - trading_halts: live halts with the reason in plain English - company_profile / check_before_paying / email_domain_check: "is this company real and who are they" before an agent pays or emails someone - a few others: read a URL as clean markdown, a market report on what other x402 sellers charge, etc. Same tools are also available as an MCP server, and they're listed in the x402 discovery index and several MCP directories. **Results after ~2 weeks (honest numbers)** - ~3,700 requests from AI crawlers (Meta, Amazon, Claude, Perplexity, Bing...) - ~1,400 MCP connections, but almost all are directory scanners and health-check bots, not real usage - ~50 calls from what look like real agents - 1 outside paying customer: an agent wallet that was buying from about 20 different x402 sellers that day. It paid for 5 of my tools on day one and came back the next morning for 2 more - Outside revenue: about 7 cents So: tiny, but not zero, and the one buyer came back, which matters more to me than the amount. **What I learned** Most "traffic" is bots indexing other bots. You need to filter hard or you'll fool yourself. Agents shop on price and coverage. The seller my buyer used most has 11,000+ listings at $0.01. I can't win on breadth, so I'm trying to win on answers that are actually correct. Feedback beats marketing. After my last post here, two people pointed out that "no filings found" is dangerous if the tool doesn't also say what window it searched and whether every source actually answered. A trading agent will treat silence as a signal. I shipped that fix the same day: the tool now says "unexplained by primary sources" instead of letting the agent guess, reports when each source was last checked, and says "halt status unknown" if the halt feed is down instead of "no halt". **How I monitor it** - A stats page that separates AI crawlers, directory scanners, real agent calls, people using the website, and my own tests. Without that split the numbers are meaningless. - A buyers table built from the payment itself: wallet, which tools, how many days it came back. - A failed-payments log, because I saw agents try to pay and fail and wanted to know why (the agent gets nothing if payment doesn't settle). - A phone push notification when a new wallet pays for the first time, or a known one comes back on a new day. - A daily morning summary of what changed. **What's next** - Add a second payment rail (RLUSD on the XRP Ledger) so agents that don't hold USDC on Base can still pay - Get listed in more directories that agents actually search - Keep every tool paid for at least once a month, because discovery listings drop tools that go 30 days without a sale - Keep improving the tools the buyer actually used rather than adding more tools **Questions for anyone doing something similar** - If you run an agent that pays for tools, how does it pick a seller? Price, description, past success? - Is anyone seeing real (non-test) x402 volume yet, or is it still mostly people buying from themselves? - What tool would you actually let your agent spend money on without asking you first?
A publishing company details how it cut daily AI agent API costs from $5.86 to $1.21 (a 79% reduction) by reducing agent wake-ups and switching from timer-based heartbeats to on-demand execution, while sharing lessons on caching, cost-per-result measurement, and moving agents off a laptop onto a server.
An AI agent with a budget and bug sent brutally honest critiques to SaaS founders, resulting in a 62% reply rate and $600 revenue, highlighting the effectiveness of utility over politeness in agentic systems.
An AI sales agent called Claw turns a website URL into a complete outbound pipeline, automatically identifying ideal customer profiles, finding companies with buying signals, and generating personalized outreach.
A Canadian real-estate agent with 60k+ stale CRM leads got a text and voice AI system built for $5k, which re-engaged old leads, flagged ready buyers, and produced 34 closed deals worth ~$490k in commission over seven months. The post details outreach numbers, consent handling, follow-up scheduling, and lessons learned.
The author shares lessons from instrumenting AI agent tool calls, revealing that tools like web_search can account for ~50% of spend, and highlighting the importance of tracking p95 latency and attributing costs per workflow or customer to avoid surprises.