10 September was a good day to watch two companies approach the same industry from opposite ends. OpenAI launched ChatGPT for Financial Services. Anthropic published a customer story about T. Rowe Price. Both aim at investment professionals. Almost nothing else about them matches.
What OpenAI built
ChatGPT for Financial Services is a product, not a prompt pack. The core is licensed data: Daloopa, PitchBook, LSEG News and Crunchbase sit inside it. Firms that already subscribe to S&P Capital IQ, LSEG, MSCI, Dow Jones Factiva or Moody’s can connect those too. On top of that come 50-plus further connectors over MCP, including Datasite, Box, Preqin, FactSet and Intapp.
GPT-6 Astra handles the research, the modelling and the client material. Every figure traces back to its original source, results arrive as interactive charts or drop straight into the firm’s own Excel, Word and PowerPoint templates. For IT there is SAML SSO, SCIM provisioning and role-based access.
Morgan Stanley and Evercore are the design partners. Morgan Stanley is quoted saying the promise of frontier research becomes real when it helps their people do the work that matters. No pricing, and availability runs through sales to “eligible financial institutions”.
What Anthropic showed
There is no Anthropic finance product at T. Rowe Price. There is Claude Enterprise, Cowork and Claude Code, and the asset manager assembles what it needs from them.
The post names three areas. Investment professionals use Cowork to pull complex material together so more time goes into analysis and debate. Teams run multi-step, knowledge-heavy processes through it where consistency and oversight matter. And developers use Claude Code to build tools for the investment and operating teams.
Behind that sits an organisational decision from August: T. Rowe Price put AI leaders directly inside Investments and Global Distribution, with T. Rowe Price Labs evaluating and scaling what works. Eric Veiel, President and Chief Investment Officer, puts it this way: “Claude lets our investment professionals cover more ground and go deeper on what matters.”
Two bets, and both can pay off
OpenAI is betting on the vertical. License the data, build the compliance features, support the templates, hand it to sales. With Morgan Stanley as a design partner you know the workflows precisely and can encode them. That is expensive, hard to copy and comfortable for the buyer.
Anthropic is betting on general tools plus the team on the customer’s side. That does not scale through data licenses; it scales when companies build things themselves. At T. Rowe Price the evidence for that is not the software, it is the org chart.
One detail I find more interesting than either launch: over 50 of OpenAI’s connectors run on MCP - the protocol Anthropic developed and opened up. The standard is winning regardless of whose product gets sold.
What it means for you
If you work in a regulated industry, this is where the decision flips: do you buy a vertical with the data baked in, or build your own on general-purpose tools? The first arrives faster. The second is yours.
Both come with a condition, honestly. The bought product ties you to your vendor’s data contracts. Building it yourself needs people who can - and T. Rowe Price created a whole structure for exactly those people. If you do not have them, you end up buying anyway.
Sources: OpenAI: Introducing ChatGPT for Financial Services, Anthropic: T. Rowe Price brings more of Claude to its investment process