Fee Watch
What exchanges changed in their market data fees and policies, read by AI from the venues' own pages, with every source linked.
A watch, refreshed every three days, over exchange market data notices, fee schedules and SEC rule filings. AgentCore Web Search finds them on the venues' own websites; a model sorts out which ones change what firms pay or how they may use data, and flags anything about AI or automated use.
How this was built
Exchanges change their market data fees and policies several times a year: new non-display categories, new licences, fee increases, new rules for derived data. Each change arrives as a notice on the venue's website or a rule filing with the SEC, and somebody at every data consumer is supposed to read them all. When nobody does, the cost shows up years later as an audit finding, like the AI agents in Audit Day that ran for months under a fee category nobody had noticed.
This lab does the reading. Every three days, Amazon Bedrock AgentCore Web Search looks for market data notices on the venues' own websites and in SEC rule filings, two searches per source: one for pages dated in the last 120 days, and one without a date limit, because many fee schedules are PDFs with no date the search index knows. A model reads each result's extract and decides whether it changes what firms pay or how they may use data, what it affects (display, non-display, redistribution, derived data, AI or automated use), and when it takes effect. A second pass writes the briefing.
Every finding links to the page it came from; the model only summarises what the extract says, so check the source before acting on it. Dates come from the page itself when it states one; the search index's date can be when a page last changed. The venues covered are the ones the search index covers well today: CME Group, Nasdaq, NYSE and ICE, Cboe, and SEC rule filings.
Exploring AI for market data policy monitoring, November 2026. Runs on Mistral Large 3 in Amazon Bedrock.
A personal project, built on a personal AWS account using personal equipment. Not affiliated with my employer.
This demo is interactive and needs JavaScript.