About the site, and me
I’m Wes. I built this site as a home for the tools listed on the front page, and as a place to work through a question I find genuinely interesting. AI agents have started paying for things on their own, and I wanted to read the receipts: for patterns, for usable information, and for the occasional oddity.
I studied quantitative economics at Cal Poly, with a minor in statistics. I started out at a couple of financial advisory firms, where I earned my Series 7 and 66, and then moved to early-stage software to see how products and companies are actually built: customer interviews, go-to-market, a churn model, a product relaunch, and one very memorable first sale. These days I consult on analytics projects and build research tools in my spare time.
I’m drawn to long, ambiguous problems, and agent payments is one of them. Most of what gets said about it concerns volume, and most of that volume turns out to be agents paying themselves. So this site counts what is left once the loops, the self-dealing and the tests are taken out (about half, as of this week), and every figure states what it is a share of. A percentage without a denominator is just a mood.
The question
When software can pay for things on its own, what does it actually buy, and what is it worth? Three smaller questions follow from that one, and each part of the site answers one of them:
- How much of the money is real? Agent payments are easy to fake: a wallet can pay itself, or a hundred wallets can be funded from one. The market, seller and buyer pages count what remains after removing those.
- What does a call cost? Listings post a price per call. Prices tracks those prices and how they change, and the endpoint monitor checks whether the posted price is the one actually charged.
- Who lets machines in? For agents to buy content, sites have to let them through. The AI-access census and the AI-policy checker record what the web’s largest sites tell AI crawlers.
What is here
Independent, reproducible data on what AI agents buy and pay for, and a few tools built on it:
- the demand-cleaned x402 index and its market dashboard;
- seller, buyer and price statistics;
- a daily endpoint monitor and a weekly census of AI-crawler access;
- tools such as Best execution, Price comps and the AI-policy checker.
x402scan is the raw explorer. This site is the part after cleaning: what survives, what buyers actually pay, and who comes back for more.
It’s built with Quarto from a public repository, using only public data: Base chain transactions via Blockscout and the Coinbase CDP Bazaar listing. The method note lists every filter and every bias I know about. If you find one I don’t, I’d like to hear it.
My own payments are excluded. I run a paid API for the index at api.wknipe.com. Payments to it, and from my test wallet, are tagged and never counted as demand. Grading my own homework seemed like a bad look.
My crawlers. The endpoint monitor (wknipe-x402-status/1.0) sends at most one unpaid request per listed endpoint per day, and no more than 60 per host. The access census (wknipe-policy-check/1.0) reads four small files per domain once a week. Both say who they are. To be left out, email wes@wknipe.com.
Privacy. Visits are counted with Cloudflare Web Analytics: no cookies, no personal data, no cross-site tracking. The AI-policy checker keeps each domain’s result for 24 hours and logs nothing about who asked.
Résumé
Experience
- Analytics and modeling engagements through Guidepoint.
- Built Opportunity Atlas, a multi-model AI research pipeline that stress-tests business models. It uses backtests, calibration checks and cross-model audits to catch AI output that is confident and wrong.
- Built this site: the x402 Clean Index, its data pipeline, API and tools.
- Business development for a BI and AI analytics platform, working between marketing and sales.
- Commissioned by the operations team to build a statistical customer-churn prediction model (logistic regression).
- Worked with operations to improve the BDR workflow.
- Interviewed current and prospective customers to shape marketing strategy, the sales motion and product direction, alongside the founder.
- Closed the first sale of the relaunched Version 3 product.
- Earned the SIE, Series 7 and Series 66 in two months.
Education
Minor in Statistics. Departmental recognition in Financial Accounting. Investments coursework built on the CFA curriculum.
Licenses and skills
- Licenses: FINRA Series 7, NASAA Series 66 (2023–2024) · FAA Part 107 Remote Pilot
- Technical: Python, R, SAS, JMP, Dynamics 365 Business Central, LLM tooling and automation
- Other: Organized Silicon Valley networking events (30–400 attendees) for technologists, investors and family offices
Contact
wes@wknipe.com · GitHub · RSS