Trading data,raw and derived.

Ask in Claude or ChatGPT, wherever you are.

Connect inmoney-data and your AI can read 12 research areas, 557 data items →

• In your own AI

  Verifying internally · opening soon

  Did insiders sell AMD this month?

  Calling get_insider_transactions

  • get_insider_transactionsinmoney-data

  Yes. Three insiders sold 41,200 shares this month, the largest sale 25,000; none bought.inmoney-data Why they sold, the data can’t say.

  Your AI sends

  get_insider_transactions

  { "symbol": "AMD", "from": "2026-10-01" }

  inmoney-data returns

  Three records, with the time they were read

  {
    "transactions": [
      { "date": "2026-10-03", "shares": 6000, "acquisition_or_disposal": "D" },
      { "date": "2026-10-08", "shares": 25000, "acquisition_or_disposal": "D" },
      { "date": "2026-10-13", "shares": 10200, "acquisition_or_disposal": "D" }
    ],
    "read_at": "2026-10-16T10:12:04Z"
  }

• Less to read

  Only the figures it needs.

  One read returns the few figures the question needs, not a web page or a whole annual report. Fewer tokens, faster answers.

• Dated

  Every figure carries its time.

  The source and read time come back with each figure. A figure on a web page often doesn’t say which day it’s from.

• Unstructured data

  The words, not just the numbers.

  Risk factors and MD&A from 10-Ks and 10-Qs, what changed since last quarter, and whole earnings calls by speaker. Searchable, in the original words.

• Fixed definitions

  Nothing pieced together.

  Every data item has a fixed name and definition, so your AI reads the same thing each time instead of stitching together sites that define it differently.

• Computed

  Ratios and ranks come back done.

  Growth, ratios, percentiles and ranks are computed from the raw data, marked as computed and traceable to their inputs. Your AI doesn’t do the arithmetic.