Trader Decision Intelligence
Make judgment visible.
Context, conflicting evidence, uncertainty and the decision to stop or ask a person.
Market judgment. Reviewable systems.
Expert-led engagements · Research workspace preview
CatalystDesk connects Sean Wu's two decades of U.S. equities experience with trader decision intelligence, institutional execution workflows and LLM agent development. We help product teams and independent builders turn complex market problems into clearer requirements, realistic evaluations and software they can test and extend.
Evidence
Market conditions
Operating constraints
Proceed within scope.
Wait for evidence.
Hand control to a person.
Illustrative design questions · A method for building and evaluating a workflow, not a trading recommendation.
Built for people building financial AI
The combination
Bring the decisions, exceptions and practical constraints into the product before they become surprises.
Make judgment visible.
Context, conflicting evidence, uncertainty and the decision to stop or ask a person.
Design for the exceptions.
Order lifecycles, partial fills, pending cancellations, reconciliation and human control.
Test the behavior that matters.
Structured context, tools and evaluation scenarios with explicit acceptance criteria.
Give the next release a boundary.
Turn a research idea into a scoped application your team can test and extend.
Two ways to begin
Product experience
Explore public context and the workspace preview. Follow the access and environment labels for analysis, simulation and testnet capabilities.
Expert services
Define a financial AI product, evaluate its reasoning, specify an execution workflow or build a bounded prototype.
Four scoped engagements
Clear outputs, defined review points and an agreed boundary for every project.
Define what to build, who it serves and how the first release will be judged.
From $3,000 USD
Explore scopeMake market reasoning explicit and test how an agent handles uncertainty.
From $5,000 USD
Explore scopeTurn institutional operating requirements into specifications engineers can implement.
From $6,000 USD
Explore scopeBuild one bounded application and collect evidence for the next product decision.
From $10,000 USD
Explore scopeIndicative 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.
Why CatalystDesk
A financial agent may summarize a market event and still miss what changes the decision. An execution workflow may look complete until an order partially fills, a cancellation remains pending or a person needs to take control. CatalystDesk brings experience from these situations into the design and evaluation of financial AI.
Sean Wu began working in U.S. equities in 2006. His experience across trading decisions, professional execution environments and systems design informs CatalystDesk’s work with AI builders today.
Meet the perspective behind the workLong-term product direction
Expert-led engagements help identify recurring needs; reusable decision knowledge, evaluation tools and workflow software are the longer-term product direction.
Future direction, not a claim that a complete institutional execution platform is available today. Client data, strategies, code and identifiable results are not reused or used for model training without written permission.
Start with the difficult part
Bring a non-confidential brief. We will explore fit, then define a bounded engagement together.
Complimentary 30-minute fit discussion.
No wallet or trading account required.
Public research
Delayed observations for research. Feed availability is shown below; these rows are not order instructions.
The public tape is empty right now. Realtime readings are on the Live Watchlist in Desk Alpha – Personal.
Loading public events…
EVENT DETAILSThe tape shows published events in the current window. Explore the Premarket Brief and Event Radar to continue your research.
Explore the research →