• Draft AI decision trees & constraints (signal priority, risk guardrails, “do-not-infer” rules) • Maintain a living playbook for how AI should interpret different market data: funding/OI regimes, unlock calendars, liquidity pockets, whale flow, basis/futures dynamics • Design workflows & product components (portfolio viewer, OI/funding monitors, holders/flow dashboards, catalyst trackers, trading reports for hot tokens) • Write clear system guidelines for how the AI condenses, cites, and interprets market data • Create evaluation rubrics & task batteries to score AI outputs (accuracy, calibration, usefulness, abstentions) • Conduct user interviews, synthesize insights, and turn feedback into shippable changes • Instrument lightweight feedback (thumbs, rationale capture, “what should it have done?” prompts) and track weekly deltas • Collaborate with engineers on offline evals + online A/Bs; iterate based on data and user signals to improve AI system’s delivery of high-quality actionable financial insights