How we are building the infrastructure for frontier AI

Jo Goodwin

How we are building the infrastructure for frontier AI

Private equity and investment banking teams are not short of AI tools anymore. Most firms already have access to Claude, ChatGPT, Copilot, or all three.

What separates the firms getting real value from the ones still piloting is not which tool they picked, it is whether that tool can see enough of the firm's own context to be useful, and whether it can be shipped to a team quickly enough to matter.

That is the thread running through everything we have worked on over the past few months. I sat down with James Ede, VP of Customers, and Martin Pomeroy, Chief Product Officer, to talk through what has actually changed, what clients are doing with it, and where it goes next.

Shipping faster, not just shipping more

Martin, what has actually changed recently, and what mattered most?

Martin: The biggest shift is not a single feature, it is how fast we can now move. We changed how we build and ship the frontend, so an idea we discuss with a client on a Monday can realistically be in their hands by the end of that same week. The new Agent Dashboard is the first thing clients will actually see come out of that, but the real change is underneath it.

James, from where you sit with clients, what has landed fastest, and what took longer?

James: The dashboard is the visible one, but the thing clients lean into fastest once they try it is the sourcing agents. It does not need a change of habit, it just starts surfacing things a person would otherwise have to go looking for. The bigger shifts, like how quickly we can now turn around a request, take longer to notice because it shows up as a pattern over weeks rather than a single moment.

A client story: value before onboarding is even finished

Talk us through a client who has gone from switched on to genuinely relying on this. What changed for them?

James: There is a client currently going through onboarding, so their process is not even mature yet, and the sourcing agents have already taken 2 opportunities to IC. That is the pattern we see most often. A client does not wait until everything is configured perfectly, the value shows up early enough to change how a team works before onboarding is even finished.

Rethinking the data layer

Martin, what is changing on the data side?

Martin: We are moving from a fixed list of around 30 connectors to single click access to the 5,000 plus connectors sitting across the growing MCP marketplaces. The other half of it is being smarter about when we spend on enrichment, so a firm can bring in far more third party context without the cost of a full data subscription for each source. It is less about having more data and more about only paying for the data that actually gets used.

The throughline

Is there a single thing tying the dashboard speed, the data strategy, and the work on frontier LLMs together?

Martin: Yes, all of it is really about building the infrastructure for a world where a firm's team is working alongside Claude, ChatGPT, and similar tools every day, not occasionally. Shipping faster matters because that world is moving quickly. The connector work matters because those tools are only as good as the context you can give them. And the work we are doing on operationalizing those tools firmwide is about making sure that using them well is not left to chance.

Where to start

What should clients who have not tried this yet go and do first?

James: Turn on the sourcing agents. It is the lowest effort thing to try and the value shows up fastest, since it starts working in the background without anyone significantly changing how they work day to day.

What is next

Martin: More on the frontier LLM side. We want teams to be able to use whichever tool they already prefer, Claude, ChatGPT, Copilot, without losing visibility or control centrally. Nothing I can put a date on yet, but that is the direction.

None of this is really about any one release. It is about a firm's context getting more useful the longer it works with the platform, and about making sure the pace of change on our side never becomes the thing holding a client back. If you want to see any of this in practice, get in touch for a demo.

Don't miss perspectives like this

Sign up to our mailing list to get insights on tech, data and AI for dealmaking efficiency in private equity, corporate finance and M&A markets.

Your AI is only as powerful as the context behind it

Stop layering tools on fragmented data. Turn market data and your own proprietary information into structured advantage to source, prioritize and surface opportunities aligned to your investment thesis.

Be first to every deal.

See Deal Engine in action.

Discover how Deal Engine is providing private equity firms with the data engineering and AI capabilities fueling their competitive advantage.