Step 1
Agree the scope
We confirm what is in scope, and that we are independent of the people who built it.
Forward deployed teamForward deployedGovernance reviewGovernanceCost reviewCost
An independent review against 32 controls in 8 domains. One result that you can show a board, an auditor or a regulator.
What you get
You receive a report, a plan with owners and dates, and a certificate of review. Every score rests on evidence.
A score for each of the 32 controls, with the evidence behind it.
Ranked by the harm each could cause. Each one has an owner and a date.
A second test, after its due date, shows the gap is closed.
Instead ofA self-assessmentAssurances by emailAnswers nobody signed
Check your own platform
Tick each one that your platform owner could show today.
Shown today, of ten
The eight domains
Who approves a change, and is that written down?
Does each business object exist once, and how does it change?
Who can sign in, and what can contractors do in production?
Is sensitive data marked where it enters the platform?
Does every automated message leave a record? Can it be stopped?
Which outside systems can the platform reach and write to?
Are time targets defined and measured?
Do audit logs reach security monitoring?
How it runs
Step 1
We confirm what is in scope, and that we are independent of the people who built it.
Step 2
We ask, look, test and collect evidence. Your own staff run any test that changes state.
Step 3
A second reviewer checks every score against its evidence. We present the result first, then the findings.
Step 4
We test each condition again after its due date.
Where it fits
The CIO, the chief data officer or the platform owner of an organization that runs Foundry in production.
Before an audit or a renewal.
Before a second ontology, the model of your business objects.
Before the first automation that acts by itself.
When the vendor's engineers hand the platform to your own team.
Read-only access for two reviewers. Interviews with the people who run the platform. Documents that already exist.
A dated list of conditions. The reviewers do not design the fix, so the list is work for your delivery team.
Nobody reviews work that they built. We certify no platform that we, or a firm we work under, built or run.
Read only. We change nothing on the platform.
Evidence is graded. A top score needs evidence that we saw ourselves.
No waiting. A live exposure is reported on the day we find it.
Not an ISO 27001, SOC 2 or regulatory certification, and not a replacement for one.
Not a penetration test.
Not a review of the platform vendor's own infrastructure.
Questions people ask
No. It is a review against a written standard, and nothing on the platform changes.
No. The review changes nothing on the platform. It needs about ten interviews, and documents that already exist.
They build and extend it. This is an independent check of what already runs, by people who did not build it.
Two Quant8 reviewers, and the second checks every score against its evidence. Most of our engineers came from Palantir, which is why we know the platform. Quant8 is independent of it, which is why you can rely on what we find.
Yes. AIP is the set of AI tools that Palantir adds to Foundry. When it runs in production, the review adds tests for it: which models are enabled, what each agent can read, and whether an agent can change data or send anything without a person approving it.
Your delivery team. The reviewers design no fix: a finding says what must be true, not how to build it. We test each condition again after its due date.
You can, and we hand over the method. With us you gain time, and a result that is independent of the team that built the platform.
You hear on the day we find it. A live exposure does not wait for the report.
Tell us the decision you must defend this year, or the Foundry platform you want an answer on. We start there.
Book a briefing James replies to you himself.