Domain Overview
GESA is Domain-Agnostic
The core GESA loop is identical across all domains. Only the episode schema and generator prompts change. The annealing mathematics, retrieval similarity function, and observable output format are universal.
The Six Domains
| Domain | Episode Source | Generator Output | Success Metric |
|---|---|---|---|
| Content | Video performance, gap measurements | Next content strategy | View count, retention |
| Workplace | HEAT signals, sprint outcomes | Team intervention | Burnout reduction, velocity |
| Trading | Fetch decisions, market outcomes | Position/timing adjustment | R:R ratio, win rate |
| Browser Automation | Fetch actions, DOM outcomes | Next automation strategy | Task completion rate |
| Business (6D) | Cascade analyses, intervention outcomes | Dimension-specific strategy | Gap closure rate |
| Product | User behaviour, conversion events | Feature/UX hypothesis | Conversion, retention |
Content
Episode source: Individual content pieces with DRIFT scores (methodology vs performance gap), view counts, retention rates, and engagement velocity (Chirp).
What GESA learns: Which content formats close the gap fastest in which audience contexts. Which Chirp signals reliably predict high engagement. What DRIFT magnitude is recoverable vs terminal for a given content type.
Temperature dynamics: Content domains move quickly — a format that worked six months ago may be saturated. Use Standard or Fast Cool profiles; apply 30-day temporal decay.
Workplace
Episode source: HEAT pain streaks, sprint velocity data, bus factor measurements, context switching frequency.
What GESA learns: Which interventions resolve pain streaks for which team profiles. How sprint phase affects intervention effectiveness. Whether WIP reduction or meeting reduction is more effective for a given team's cognitive load pattern.
Temperature dynamics: Team dynamics change slowly. Use Standard Cool; apply 180-day temporal decay. New teams start at Slow Cool.
Full integration: GESA × HEAT →
Trading
Episode source: Fetch decisions with market conditions (Chirp = signal strength, Perch = structural setup confidence, Wake = temporal pattern context), entry/exit outcomes.
What GESA learns: Which Fetch threshold produces the best R:R ratio for which market condition profile. Whether patience (Wait) or action (Execute) produces better outcomes in a specific volatility regime. Optimal cooling schedule for market cycle awareness.
Temperature dynamics: Highly variable by market regime. Adaptive Cool is the preferred profile — let episode outcome variance drive the cooling rate.
Browser Automation
Episode source: Fetch decisions against DOM structures, selector success/failure rates, page navigation outcomes, task completion rates.
What GESA learns: Which selector strategies are most robust across page structure variations. When to retry vs escalate on failed actions. Which DOM patterns reliably indicate a specific page type.
Temperature dynamics: Page structures change frequently — old episodes may be stale. Use Fast Cool with 14-day temporal decay. Stale episodes should be purged aggressively.
Business (6D)
Episode source: 6D cascade analyses (FETCH scores, dimension origins, cascade paths), intervention outcomes measured by gap closure rate.
What GESA learns: Which cascade patterns are most predictive of downstream impact. Which intervention types close D3 Revenue gaps vs D6 Operational gaps. How cascade depth affects the effectiveness of different intervention strategies.
Temperature dynamics: Business strategy domains move slowly. Use Slow Cool with 180-day decay. Allow thorough exploration before converging on intervention patterns.
Product
Episode source: Feature releases with conversion and retention outcomes, A/B test results, user behaviour segmented by user type and lifecycle stage.
What GESA learns: Which feature hypothesis types produce conversion lifts for which user segments. Whether to prioritise acquisition (new user features) or retention (existing user features) given current gap trajectory. How long to wait before attributing an outcome to a product change.
Temperature dynamics: Product domains have medium velocity — seasonal patterns matter. Use Standard Cool with 90-day temporal decay.
Adding a New Domain
GESA's domain-agnostic architecture means new domains can be added by:
- Defining the episode schema — what context, action, and outcome fields are specific to this domain
- Specifying the success metric — what does "gap closed" mean in this domain?
- Setting the decay constant (τ) — how quickly do historical episodes become stale?
- Choosing the initial temperature profile — how well-understood is this domain?
- Implementing the generator — rule-based, LLM-prompted, or hybrid
The core retrieval, annealing, and output layers require no modification.