Public CatalystDesk service and product information.

Source: https://catalystdesk.ai/en/services/trader-decision-intelligence

[Services](/en/services)Decision Intelligence & Evaluation

Scoped expert engagement

# Trader Decision Intelligence & Agent Evaluation

Make market reasoning explicit and test how an agent handles uncertainty.

[Request a Scoped Proposal](/en/contact?intent=proposal&service=trader-decision-intelligence)

Choose a starting scope

## Focus, Team or Extended.

Financial AI teams and agent developers who need realistic decision scenarios and traceable evaluation evidence.

Focus

### $5,000 USD

Estimated 2 weeks

- One workflow and up to 20 scenarios.
- One frozen agent configuration.
- One evaluation cycle.

[Discuss Focus scope](/en/contact?intent=proposal&service=trader-decision-intelligence&tier=Focus)

Team

### $10,000 USD

Estimated 3–4 weeks

- One workflow and up to 50 scenarios.
- Two frozen configurations and two evaluation cycles.
- Remediation review.

[Discuss Team scope](/en/contact?intent=proposal&service=trader-decision-intelligence&tier=Team)

Extended

### $18,000 USD

Estimated 5–6 weeks

- Up to two related workflows and up to 100 scenarios.
- Up to three frozen configurations and two evaluation cycles.
- Lightweight evaluation runner for the agreed input/output format and handover.

[Discuss Extended scope](/en/contact?intent=proposal&service=trader-decision-intelligence&tier=Extended)

Indicative fees in USD for the defined scope. Final scope, timing and commercial terms are confirmed in a statement of work. Estimated delivery periods begin after scope, usable inputs, authorized access and the kickoff payment are ready.

What you receive

## Work your team can review and use.

- Structured decision framework, including uncertainty, conflicting evidence and human escalation.
- Scenario pack with inputs, expected behavior, scoring criteria and failure categories.
- Traceable scored results and prioritized improvements.
- Extended: a lightweight evaluation runner for the agreed input/output format.

Acceptance

### Define done before kickoff.

Acceptance concerns the completeness and quality of the evaluation, including traceable results against the agreed criteria. A low score can be a valid finding. Model improvement and trading profitability are not guaranteed.

Final deliverables, acceptance criteria, responsibilities and dates are agreed in the statement of work.

Client inputs

## What we need to begin.

- The workflow, intended behavior and an agreed model-and-prompt or agent configuration.
- Agreed public, synthetic or lawfully client-authorized materials.

Scope and limits

## Keep the boundary clear.

- A scenario is one documented situation with its inputs, expected behavior and judgment criteria.
- A configuration is a frozen model and prompt or agent setup. Each evaluation cycle covers the agreed scenarios and configurations.
- One consolidated document revision is included. Extensive labeling, additional data procurement and agent remediation development are separately scoped.
- This service does not provide a comprehensive historical market database or a certified benchmark.

Commercial details

## No surprises between scope and handover.

Fees exclude taxes and third-party costs such as data, model/API usage and cloud services. Providers and spending caps are agreed in advance. Changes to scope require written agreement.

[Read the Engagement Process](/en/how-we-work)

Start with the difficult part

## What does your financial AI need to get right?

Bring a non-confidential brief. We will explore fit, then define a bounded engagement together.

[Request a Scoped Proposal](/en/contact?intent=proposal)

Complimentary 30-minute fit discussion. No wallet or trading account required.