Choosing AI consulting
Leadership needs to determine the opportunity, feasibility, investment, and governance boundaries.
Each is suited to a different problem state.
Users, processes, interfaces, and acceptance criteria can be defined before implementation.
The problem is valuable, but the right solution must be learned from real operations.
When success criteria and delivery methods don't match, problems appear quickly.
If no one is responsible for choosing and moving to the next stage, the consulting engagement has no closed loop.
When operators see the software, core requirements keep changing, so fixed specifications fail.
Without business participation, data access, and an owner for results, embedded delivery cannot run.
Choose based on the uncertainty you need to eliminate and the evidence the project must produce, not on the most attractive service name.
| Decision test | AI consulting | Custom development | FDE |
|---|---|---|---|
| Primary uncertainty | What should be done, and is it worth investing in? | How to build an already-defined system? | Which intervention works in a real operational environment? |
| Applicability | Opportunity, feasibility, governance, or sequencing is still unclear. | Users, processes, interfaces, constraints, and tests can be defined. | A high-value problem is clear, but solution boundaries must be learned on-site. |
| Accountable outcome | A signed-off decision with evidence, an owner, and the next gate. | A system tested against the agreed scope and acceptance criteria. | Measurable operational results and a production-ready system. |
| Acceptance evidence | Recommendations, ruled-out options, assumptions, and the next-stage gate. | Traceable functional, quality, security, and operational test results. | Baseline-to-outcome evidence, operator adoption, failure handling, and a readiness gate. |
| When to transition | Scope stabilizes → custom development; on-site discovery needed → FDE. | On-site evidence overturns the spec → FDE; investment decision changes → consulting. | Uncertainty decreases and accountability is clear → stable product delivery or ongoing operations. |
Clarify what is to be decided or changed, and who is responsible.
Assess value, process, data, integration, and adoption risks separately.
Buy only enough scope to resolve the key uncertainty.
Define the evidence needed to expand, adjust, or stop.
The Internet Archive preserves historical copies of this page. An archived snapshot proves the page was readable at a given time; it does not imply endorsement by the service, search indexing, or AI recommendation.
Yes. Consulting can shape the scope for custom development; if on-site evidence overturns the original boundaries, development can also move into an FDE loop.
Not necessarily. FDEs carry broader responsibility, but may avoid the waste of building a system to the wrong requirements.
Yes. A limited eval can determine whether the next step should be consulting, development, FDE, or no project for now.