FEATURED · DATA INFRASTRUCTURE
Half of new web articles are now AI-written and their citations often don't hold up. A trusted data layer gives agents structured, cited SEC filings to read before the noise.
Jennifer Ma
13F holdings and insider trades come over a REST API; earnings-call retrieval comes over MCP. Here's how to wire each into a Claude workflow, with real config.
Jennifer Ma
Two 2026 benchmarks show the same pattern: frontier models that rarely hallucinate on general text get measurably worse on multi-step financial calculations.
Jennifer MaBerkshire's first 13F under Greg Abel and Li Lu's Himalaya both leaned into Alphabet in Q1 2026 — here's what the filings show, and why the snapshot is stale.
Jennifer MaCoverage claims, silent substitutions, and taxonomy drift break financial agents quietly. Here's what to actually test before trusting a data vendor in production.
Jennifer MaFive vendors — financialdatasets.ai, FMP, Massive, EODHD, and FocusAlpha — call themselves financial data APIs, but cover, price, and target different things.
Jennifer Ma13F shows what institutions own; Form 4 shows what insiders buy. Overlay the two for a consensus signal — and learn the crowding trap that breaks it.
Jennifer MaA 13F shows only long, US-listed equity as of quarter-end. The four structural blind spots in institutional holdings data — and how not to over-read them.
Jennifer Ma13F Pro runs 10 AI analysts that debate 6,000+ stocks daily, every claim tied to an SEC filing. A look at its design and the build-or-buy question beneath it.
Jennifer MaA viral investor thread said the model (the Brain) barely matters — the Hand (data, tools, context) differentiates financial agents. 2026 benchmarks agree.
Jennifer Ma13F institutional holdings data reaches the public 45 to 135 days late. Here's how an AI agent layers Form 4 trades and 8-Ks to close the timeliness gap.
Jennifer MaA new study ran LLM-compressed 10-Qs and earnings calls through an investment-decision task. Summaries flipped the decision in up to half of cases — without saying anything false.
Jennifer MaThe most common question we get: won't better models make the data layer unnecessary? The benchmark evidence points the other way — the ceiling on answer quality is set by the data.
Jennifer MaWhat a SEC filings API needs before AI agents can rely on it: EDGAR's real constraints, the failure data on retrieval pipelines, and a seven-point checklist.
Jennifer MaEDGAR is free and its JSON APIs are real — so when does a SEC filings API earn its keep? An honest cost breakdown for teams wiring AI agents to SEC filings.
Jennifer Ma8-Ks are how public companies announce material events. Monitoring them with an agent fleet sounds simple — until EDGAR's rate cap, amendments, and event classification get in the way.
Jennifer MaKeyword search finds a word; agents need to find a theme across dozens of earnings calls. What an earnings call transcripts API needs to make cross-company theme retrieval work.
Jennifer MaA study ran LLM summaries of earnings-call transcripts through an investment-decision task. The summaries flipped the call in up to half of cases — usually by cutting one management qualifier.
Jennifer Ma