Determa Determa Applied AI Systems

One blocked project.
The context to act.

Your ticket says blocked. The CRM still promises a launch date. Determa connects the evidence so the project owner can review the next action and record the result.

Scope your workflow30-minute intro · free · no obligation

See the workflow

Evidence. Review. A recorded result.

The open-source starter follows a fictional project through one decision. It runs locally with a Python SDK and a browser interface.

  1. Read a blocked ticket alongside the customer commitment and an earlier decision.
  2. Draft a comment and review its exact wording with the source records.
  3. Approve and post to a local Jira simulator. Inspect the receipt.
  4. Change a cited source and see the outdated proposal refused.
Download the starter and Python SDKMIT license · Python 3.12+ · synthetic data · no account required

The starter uses a local simulator and demonstration identities. Client work adds real connectors, identity integration and the controls required for the agreed workflow. It is not a complete enterprise platform.

The first paid engagement

A decision you can act on.

After the free scoping call, commission a workflow assessment when the problem and owner are clear. You receive a written recommendation before committing to implementation.

  • One user story, its owner and measurable acceptance criteria.
  • A map of sources, permissions, decisions and permitted writes.
  • A build-or-buy recommendation, including reasons to keep the current process.
  • An implementation estimate by phase and the expected operating responsibilities.

The assessment is paid work. Scope, hourly rate and effort by phase are agreed before it starts. Live integration is a separate commitment.

If the assessment supports a build

Prove the smallest useful workflow.

Connect the agreed sources, enforce the user's access, bind approval to the evidence and test the real write path. Handover includes the source, deployment instructions, failure cases and operating responsibilities. Hosting, graph retrieval and model use are chosen for that scope.

Inspect the work

The case study describes a separate client implementation. The downloadable starter was created independently with fictional records.

Before we speak

Can our team build on the SDK itself?

Yes. The MIT-licensed source includes a Python SDK, connector example and tests. Hiring Determa is optional.

Will this replace Palantir Foundry?

A scoped context layer can serve a defined workflow. It does not replace Foundry's full platform. The assessment compares the capabilities you actually need.

Can it run on our infrastructure?

That is part of the deployment scope. Identity, data sources, hosting and any models must be agreed and verified together. The synthetic starter already runs locally without model calls.

Bring one concrete problem

Who owns it? What fails today?

Bring a workflow owner, the systems involved and an anonymized example of a missed change. The first call checks whether a bounded implementation is likely to help. If you need a shared platform across many applications, the assessment should compare platforms instead.

Scope your workflow30-minute intro · free · no obligation