Ontology design and data integration
The object model that makes Foundry useful: your deals, policies, claims, counterparties and the pipelines that keep them current. Built with the people who own the data, not just the people who own the platform.
You don't need a forty-person programme to get Foundry and AIP working. Senior Treefrog engineers lead focused, fixed-scope builds you can see working in weeks — and we have worked out how to make building on AIP dramatically faster than the out-of-the-box experience.
The object model that makes Foundry useful: your deals, policies, claims, counterparties and the pipelines that keep them current. Built with the people who own the data, not just the people who own the platform.
Front-line tools in Workshop that your underwriters, claims handlers and finance team actually open every morning, shaped by a Living Requirements Document before they are built.
Agents that summarise submissions, triage claims, draft responses and route work, with the guardrails, evaluation and human review that make them trustworthy in a regulated business.
Already on Foundry and not getting the value? We look at what is there, find what is not working, and extend it with focused builds rather than starting again.
How a Foundry build takes shape
Foundry earns its keep when the ontology matches how the business actually thinks: policies, claims, counterparties and brokers as objects, linked the way your underwriters link them in their heads. Get that right and a Workshop app or an AIP agent grounded in it becomes a matter of weeks.
The last frame is the part nobody else shows you: the same five changes to a Workshop app, made two ways. Anyone who has waited on AIP's own tooling will recognise the left-hand side.
The 30-second version
Our animated short announcing the Palantir practice. Thirty seconds, no slides.
Anyone who has built on AIP knows the wait: a simple change can take the platform's own AI tooling the better part of an hour. We have developed a way of working that turns those changes around in minutes, with the engineer still in control. It is the single biggest reason our Palantir builds land in weeks. Ask us about it on a call.
A fixed-scope first build is typically live in six to twelve weeks. Large programmes are still staffing the proposal.
Our Palantir practice launched in 2026. Every engineer on it has at least ten years of software development behind them, and they have been shipping Foundry and AIP work since the practice began. Alongside them is our founder, with 27 years building pricing, portfolio and claims systems for the (re)insurance market, who oversees the design so the ontology reflects how the business really works.
Twenty-plus years in (re)insurance, ILS and funds. The ontology we build reflects how underwriters actually think, because we already speak the language.
We build in your Foundry instance, document everything and hand over. Nothing is held hostage.
We are not a Palantir partner or reseller, so we have no reason to recommend the platform when it is the wrong tool. If it is, we'll say so and build on your stack or ours instead.
How a Palantir engagement runs
To build in Foundry you need a Foundry tenancy. If you have one, we build in it. If you are evaluating Palantir, we will tell you honestly whether it fits the problem before you commit.
No. We are an independent consultancy, not affiliated with or endorsed by Palantir, and not a reseller. We build on the platform in your tenancy and give you a straight answer about it.
A fixed-scope first build — an ontology slice plus one Workshop application or AIP agent — is typically live in six to twelve weeks, with weekly demos throughout.
Our Palantir practice launched in 2026. Every engineer on it has at least ten years of software development behind them, with design oversight from our founder, who has 27 years in (re)insurance systems. That gives every engagement engineering depth and domain depth, plus a way of working that makes AIP builds dramatically faster than the standard tooling.
Anyone who has built on AIP knows a simple change can take the platform's own AI tooling the better part of an hour. We have developed a way of working that turns the same changes around in minutes with the engineer in control. We explain it on a call rather than on a web page.
Yes. We look at what is there, find what is not working, and extend it with focused builds rather than starting again.
Thirty minutes with the team behind our Palantir work. You'll leave with a scoped first build and a straight answer on whether Foundry is the right tool for it.
Palantir, Foundry and AIP are trademarks of Palantir Technologies Inc. Treefrog Consulting is an independent consultancy and is not affiliated with, endorsed by or a partner of Palantir.