Is ChatGPT or Claude Enough For Your Fund?

Investment firms have gained a new superpower: AI built specifically for finance. Anthropic launched Claude for Financial Services in July 2025, and OpenAI followed suit this September with ChatGPT for Financial Services. Drawing on both a firm’s own data with market data, these tools can do it all: from screening deals to monitoring portfolios.

But a superpower stops being ‘super’ once everyone has it. If any fund can access the same AI tools, what gives anyone an edge over the rest?

The Benefit of AI Is Capped By Your Data

The need for AI is clear. Most investment firms run on small teams, yet each one has to sift through hundreds of documents to source deals and spot promising opportunities, all while tackling daily administrative tasks. For just a handful of people, that workload is enormous. This is where AI steps in, taking on the heavy reading and freeing investors to focus on what only people can do: meeting companies and building relationships.

But this only works if the AI output can be trusted. Out of the box, these tools come with a blind spot: they may know the market, but they don’t know your fund.

Without context on your deal history, portfolio data, and decision patterns, AI will pump out the same generic answer as the fund right next door. The obvious fix is to give AI more: upload every deck, memo and set of financials, and let it work everything out. But piling in large amounts of data can make their answers less accurate as they begin to hallucinate.

Why does this happen? It seems counter-intuitive, but there are three main reasons.

1. Attention gets spread thin. The more data you provide a model, the more the important details have to compete with irrelevant material. Models also tend to pay more attention to the start and end of a long input, so key information in the middle may get overlooked.
2. Old material gets in the way. A deal builds up a long trail of documents: earlier drafts, revised figures and corrections all remain in the view. So, outdated information can influence the latest answers.
3. Errors snowball. If there are erroneous figures due to human error, or the model makes a mistake early on, it stays in the conversation. Without any manual correction, later answers may be built upon these errors.

How Funds Can Fix Their Data

The world’s leading firms already know this. Swedish investment firm EQT has built their own AI platform, Motherbrain, which combines external market data with the firm’s internal knowledge: meeting notes, past assessments and the connections that its employees upload. All of this data is scattered across emails, decks, and spreadsheets. Motherbrain gathers and organises EQT’s data into one structured system, and run its own AI on top of it. This is what helps it make faster, better-supported decisions. OpenAI itself names “reliable access to data” as one of the biggest pain points for financial teams. The lesson is clear: use the best models available, but give them your firm’s data in a form they can actually use.

EQT spent several years developing Motherbrain, deploying a dedicated data science team just for it. Many investment teams are small and their resources are stretched thin, therefore can’t do the same. Bigger funds are attempting to build similar platforms with engineers, but they are lacking data science capabilities to fine tune their product, which they don’t have the time nor money for.

Many funds are now using Morpheus, which is the intelligence layer that sits between a fund’s data and the AI it uses. It organises everything from pitch decks to call notes into a structure specifically built for investment work. The more deals a fund works through, the more accurate the answers become, not less. When sources conflict, it knows which one takes precedence. As the team uses it, it learns from their corrections and picks up how exactly tasks are done, something off-the-shelf tools simply don’t do.

Morpheus also offers flexibility: investment firms can run Morpheus’ own skills, or plug ChatGPT, Claude or their in-house agents on top of Morpheus’ intelligence layer. The choice is yours.

AI built for finance is a superpower, but only for funds whose data is ready for it. Every firm can buy the same tools; the real edge comes from the data they feed in, and how it’s well structured for AI to use. The funds that get their data AI-ready are the ones with the winning cards.