What is AI Solution Engineering?

AI Solution Engineering is the discipline of turning ambiguous business problems into valuable, feasible, and trusted AI solutions—and leading stakeholders from discovery to an explicit decision.

It is the primary name of this playbook. The interactive tools you already use (ConsultAI OS, architecture map, FinOps, canvas) hang off this spine. The methodology that organises the work is the 8D AI Solution Engineering Framework.

Why a discipline—not just a toolkit​

Most AI initiatives fail for process reasons, not model reasons:

• A model is chosen before the problem is defined
• Workshops produce slides, not decision records
• Risk and governance arrive as end-of-project paperwork
• Executives never get a clear ask

AI Solution Engineering inverts that. Problem before AI. Output before content. Evidence over confidence.

The 8D spine​

Primary navigation uses eight dimensions. The older 20-stage ConsultAI lifecycle remains as detailed tools under each D:

Stage	Gate (you do not advance until…)
Define	Problem is expressed without naming a particular AI product or model
Discover	Evidence-backed opportunity list exists
Diagnose	Every proposed solution addresses a verified cause
Design	Target workflow and human–AI allocation documented before tech lock-in
De-risk	Owner, control, evidence, and residual risk for each material risk
Demonstrate	Go / modify / stop recommendation with evaluation evidence
Decide	Explicit ask and decision log entry
Deliver	Approved roadmap and benefits plan with owners

Full stage pages and artefacts are specified in the product docs; this post is the practitioner overview.

VALUE — the quality gate on every artefact​

Before you advance a stage or send something to an executive, run VALUE:

Letter	Question
V Valuable	Material business need?
A Actionable	Specific enough to execute?
L Logical	Structured and evidence-backed?
U Understandable	Can the audience grasp it without unnecessary effort?
E Executable	Ownership, constraints, risks, and next steps clear?

Trust is scored separately: privacy, security, fairness, transparency, human oversight, auditability, reliability.

See the full VALUE gate.

How this maps to the live playbook today​

You want…	Open today
Consulting lifecycle & workshops	ConsultAI OS (legacy 20-stage tools under 8D)
Cross-cloud architecture	Architecture Map, Compare, Canvas
Cost & alternatives	LLM FinOps
Decision trade-offs	Decision Assistant
Full curriculum	Guide overview

V1 of the product is content-led: teach 8D publicly, keep the SPA as the working toolkit, then grow authenticated workspaces in V2+.

Design principles (non-negotiable)​

1. Output before content — every page produces a reusable artefact or decision
2. Problem before AI — no model selection before Define/Diagnose gates pass
3. One framework, many tools — 8D is the spine; tools hang off stages
4. Answer first — bottom line before supporting detail
5. Evidence over confidence — label Fact / Assumption / Hypothesis
6. Trust by design — risk and governance visible throughout, not bolted on
7. Decision oriented — every engagement ends in an explicit decision record