Opportunity assessment
Test the decision, available history, authority to act and value at stake.
Qualify, prove and deploy
Fable tests the opportunity, measures the engine against an agreed baseline, deploys it inside the client boundary and remains responsible for its performance.
The process has four stages, each with its own decision gate.
Proceed, change the scope or stop. Evidence determines whether the work advances.
Pilot
Run the engine against an agreed baseline using the client’s operating record.
Production
Deploy the approved models, workflow, integrations and fallback.
Stewardship
Monitor performance, retrain when evidence warrants it and account for results.
01 · OPPORTUNITY ASSESSMENT
Establish whether the opportunity is viable.
The opportunity assessment identifies the recurring decision, its owner, the evidence available at the time, the actions under the organisation's control and the method for measuring improvement.
| Decision | A repeated operating choice with a named owner and a material consequence. |
| Evidence | Historical records that can be reconstructed as they stood before the outcome was known. |
| Output | A recommendation to proceed, change scope or stop. Viable opportunities include a projected uplift, a pilot design and a commercial case. |
| Fee | A fixed fee, agreed before the work starts. Implementation and the annual licence are then priced from the projected uplift. |
Compare the engine with current practice
Map client records, fit and calibrate the candidate models, freeze the baseline and test whether the engine improves the agreed decision.
Deploy the approved decision path
Put the selected models, fallback, operator workflow, integrations and measurement into the client environment.
Keep the engine fit for use
Monitor data, model performance, policy, completion and outcomes. Retrain or recalibrate when the evidence justifies a change.
Reusable product and client implementation.
| Fable product | Client implementation |
|---|---|
| The decision model, forecasts, action types, constraints, optimisation, value method and operator workflow. | Source access, field mapping, historical records, owners, permissions and integrations. |
| Calibrated production models, versioned model releases and a rules-based fallback. | A client-calibrated model, shadow evaluation, approval and production limits. |
| Contracts for recording decisions, actions, outcomes and financial value. | The agreed baseline, observed results, finance review and rollout. |
How decisions are checked.
Results come from your records and a baseline your finance team agrees.
- Every recommendation links to its source records, model version, alternatives and constraints.
- A model can influence a decision only after it beats simpler rules-based and statistical methods on records it has not seen.
- Your team approves each action before it changes a source system.
- The completed action and the observed outcome stay on the same record, so the value can be reconciled with your finance records.
Qualify the decision before committing to a pilot.
Tell us how it is made today, which records support it and what a better result would be worth.
Book a live walkthrough