Ayati Decision Intelligence Playbook™
A single-file knowledge system covering the foundations, architecture, methodology, platforms, and governance of Decision Intelligence. Built for workshops, client engagements, and internal capability development.
How to use this resource
The Ayati Decision Intelligence Playbook™ is a structured knowledge system covering everything you need to understand, design, and operate a Decision Intelligence capability. It is designed to be used during workshops, shared with clients, and referenced by internal teams building analytics systems.
The playbook is: single-file and self-contained (Mode-B compliant), deterministic in content — no generated or uncertain material, print-friendly for chapter-by-chapter export, and search-enabled with scroll-spy navigation.
The Decision Problem
Many organisations invest in analytics and dashboards, but the quality of decisions stays flat. The failure mode is predictable: metrics multiply, definitions drift, ownership is unclear, and insights do not translate into action loops.
Decision Intelligence shifts the unit of design from "reporting" to "decision systems": inputs → assumptions → rules / models → governance → actions → learning.
Common failure patterns
- Definition drift: KPI meaning changes informally over time
- Orphan metrics: no decision owner, no SLA, no action playbook
- Model opacity: black-box outputs without acceptance criteria
- No experiments: changes rolled out without controlled validation
- No receipts: inability to reconstruct inputs and assumptions later
Decision Intelligence is the practice of engineering repeatable, governed decision systems that convert data into actions and learning. Output quality is not a UI problem — it's a system design problem.
Ayati Philosophy
Start with decisions and owners. Analytics exists to reduce uncertainty around a specific choice — not to generate reports.
Definitions, assumptions, and model logic must be inspectable. Transparency is not a feature — it's the foundation of trust.
Experiments and learning loops prevent "insight theater." Every decision should generate evidence that improves the next one.
Systems should run reliably even in low-connectivity environments (Mode-B). Privacy by design is a governance principle, not a technical constraint.
Operational principles
- Ownership: every KPI and decision has an accountable owner
- Definitions: metrics are versioned; disputes have a resolution path
- Controls: thresholds, guardrails, and approvals reduce risk
- Auditability: every decision has "receipts" — inputs, logic, and output
- Portability: systems should run reliably even in low-connectivity environments
DI Maturity Pyramid
The maturity pyramid maps capability levels to typical engagement models. Each level represents a distinct state of Decision Intelligence capability — and a different way of working with Ayati.
Strategic Decision Architecture
Strategic decision architecture defines how an organisation structures the flow from data to decision to action. It is not about dashboards — it is about the explicit, governed system that connects metrics to outcomes.
The DI Canvas
The DI Canvas is a structured framework for mapping a decision domain. It captures: the decision being made and who owns it; the inputs and their provenance; the rules or models applied; the governance and controls; and the expected actions and learning loops.
Ecosystem Map
The ecosystem map shows how Decision Intelligence connects across the organisation — which teams produce signals, which consume decisions, and where the governance boundaries sit. It prevents the common failure of building siloed analytics capabilities that drift apart over time.
The Decision Lab
A Decision Lab is a time-boxed environment for exploring a decision domain before committing to a full build. It uses Mode-B prototypes to validate the logic, definitions, and governance structure — so platform investment is preceded by validated design.
Methodology & Visual Boards
Ayati's methodology is deterministic: every step produces a documented artifact, every choice is traceable, and every output can be reviewed and reproduced. This is not a methodology for generating reports — it is a system for engineering repeatable decision quality.
Visual Boards
Visual boards are the primary governance artifact for a Decision Intelligence engagement. They show the KPI tree, driver attribution, assumptions log, and action owners in a single view — so leadership can interrogate the logic without needing an analyst in the room.
Core methodology steps
- Define the decision: who owns it, what success looks like, what constraints apply
- Map the drivers: decompose outcomes into controllable inputs using a KPI tree
- Encode the logic: translate drivers into deterministic rules with traceable branches
- Validate: run scenarios, edge cases, and governance review before deployment
- Enable: build the training, playbooks, and confidence layer for adoption
- Govern: version definitions, log changes, and maintain audit trails over time
The Ayati Product Ecosystem
Four are agentic, LLM-powered productized services; MedStat Guide™ is a standalone offline app. Together they cover the decision lifecycle — from mix decomposition and contradiction intelligence to false-belief detection, multi-engine synthesis, AI research, and clinical statistics.
Adaptive Mix Intelligence. Decomposes your business into its decision-relevant mixes — acquisition, channel, customer, retention, funnel — and drives each from description to decision, sized in money. Productized service.
Standalone offline app — a deterministic statistical engine for clinical research. 100+ models with full rule traces, effect sizes, and publication-ready outputs.
Decision reconciliation engine. Recomputes shared figures and settles conflicting analyses with a test run in code — one audit-traceable answer.
Agentic AI research platform. Governed study-portfolio design, reasoning, and publication-ready output, audit-traceable end to end. Productized service.
Contradiction-intelligence engine. Finds where your business laws have quietly broken, scores each paradox, and prices what it is costing — on any operational dataset. Productized service.
Strategic Autonomous Generative Engine. Detects the false belief your organisation optimises around — and why it persists — sized in money, in plain owner language. Productized service.
Governance & Risk
Governance in Decision Intelligence is not a compliance exercise — it is what makes the system trustworthy enough to act on. Without governance, analytics decay: definitions drift, ownership evaporates, and decisions lose their audit trail.
What gets governed
- Metric definitions and calculation rules — versioned with a change log
- Model assumptions and limitations — declared before deployment, updated when they change
- Decision ownership — every KPI has a named owner with a resolution path for disputes
- Release notes and validation evidence — every change is documented before it ships
- Audit traces — inputs, rules, and outputs preserved for reproducibility
Analytics culture
Governance only works when it is embedded in the culture — not bolted on as a compliance layer. The goal is to make "checking the definition" and "reading the trace" habitual, not burdensome. This is the work of adoption engineering: building the confidence layer that makes governance feel like support, not friction.
Case Templates
Retail Pricing Case
A pricing decision domain for a multi-brand retail context. The case covers: revenue decomposition into traffic, conversion, and AOV; pricing lever attribution and elasticity estimation; scenario modelling for promotional trade-offs; and governance of pricing assumptions and owner assignment.
The case template includes a DI Canvas, KPI tree, assumptions log, and a worked example of a pricing intervention with full decision receipt.
Clinical Outcomes Case
An outcomes measurement domain for a clinical research context. The case covers: endpoint definition and analysis population rules; deterministic method selection with rule trace; effect size reporting and clinical interpretation; and governance of analysis decisions for peer-review defensibility.
The case template includes a study design map, method selection trace, diagnostics summary, and publication-ready output structure.
Case templates are starting points — not scripts. Adapt the DI Canvas and KPI tree to your specific decision domain. The governance structure and audit trail requirements stay consistent across all cases.
Use this playbook in a live engagement
We'll adapt these frameworks to your decisions, your data, and your team.
Talk to Ayati →