Morningstar has 1,100 analysts behind its Claude connector

Morningstar ($MORN) launched something called Direct AI earlier this month, and the release says they are making their research available through Claude, Copilot and ChatGPT, so you can reach it from whichever AI tool you already work in. The launch itself is three agents built on Morningstar's data and ratings. One is for asset managers deciding how to position or launch a fund. One is for sales teams trying to get a fund onto an approved list or into a model portfolio. The third is for due diligence teams vetting fund managers, and it uses the same framework Morningstar's own analysts use, with their categories and Medalist Ratings built into the logic and more than 1,100 analysts behind it. Scott Brown, who runs the Direct platform, put the reason in one line: "AI and agents are only as valuable as the data that powers them." It feels like every research data company has started doing some version of this now. On 14 September Anthropic also put out Claude for Financial Advisors, with connectors from BlackRock ($BLK), Schwab ($SCHW) and Vanguard among others, and Schwab says it is bringing it to the 16,000-plus independent advisers on its platform. Morningstar shows up on that list too, as one of the connectors that already existed. The Morningstar connector itself has been in Anthropic's directory since November 2025. It lists six tools, which look up the investment you mean, pull data on funds, ETFs and stocks, pull excerpts from their analyst research, pull their articles, run a screener, and show a fund's top holdings. You sign in with a Morningstar account, so for now it is built for people who already have Morningstar access. Two things I still could not work out, and maybe someone here knows. If your firm has turned this on, when you ask Claude about a fund, does the data tool hand back Morningstar's actual numbers, or mostly the rating and a summary of the analyst's view? And for those of you already doing investment research in Claude, would you pay for a vendor's data inside it, or do you prefer pulling from free sources like EDGAR yourself?