Do I need an AI idea before starting?

No. Start with a process, its outcome and an observed problem. The map helps you identify and compare candidate interventions.

Can the recommendation be a process fix rather than AI?

Yes. Process repair, integration and rules are legitimate alternatives. AI is useful when its capability addresses the constraint and performs reliably enough in representative tests.

Are recommendations or ROI validated predictions?

No. Decisions depend on your inputs, disclosed assumptions and delivery conditions. Replace estimates with operational evidence. The tool does not provide peer benchmarks.

Do released hours count as cash savings?

Released hours are capacity. Cash savings need an identified cost reduction, avoided hiring or another credible realization path. Review and rework must be included in net time saved.

Can I compare several processes?

Yes. Keep the outcomes, evidence and assumptions visible. Identify shared enabling work and avoid counting the same benefits in multiple opportunities.

AI Opportunity Map

By Juan Beltrán, Industrial B2B AI Transformation Executive.

Start with a process and its performance gap. Compare process repair, integration, rules, predictive models and generative AI, then build a shortlist with assumptions, missing evidence and the next experiment. Recommendations are transparent decision aids for discussion with process owners.

What it looks at

Process discovery without needing a preselected AI idea
Intervention alternatives with their expected mechanism
Evidence provenance and explicit assumptions
Benefit ranges with explicit cost and adoption assumptions
Separate deployment conditions and enabling work
A shortlist and a concrete next experiment

Who it is for

Process owners, operations teams and transformation leaders who know where work is slow, costly or unreliable and want to identify which interventions deserve investigation. Industrial B2B examples cover quoting, fulfillment, service, tenders, onboarding and planning.

How the score is constructed

Five decision areas: process and outcome, constraint and evidence, intervention alternatives, delivery conditions, and economics. The method draws on APQC process guidance, task-level AI research and NIST AI RMF principles. It is a practical synthesis, not an empirically validated ranking model. Evidence confidence is separate from opportunity potential; deployment blockers cannot be averaged away.

Process and outcome

Define the process from its trigger to the result delivered to a customer. Name an owner, the current performance gap and the outcome worth improving.

Constraint and evidence

Locate waiting, rework, exceptions and handoffs. Distinguish knowledge or interpretation problems from missing capacity, unclear authority, policy or incentives. Record competing explanations.

Intervention and alternatives

Compare process repair, integration, deterministic rules, predictive models and generative AI. Describe the mechanism linking each intervention to the outcome. Combinations are allowed.

Delivery conditions

Check input access, acceptable error, evaluation, human review, action authority, integration and adoption. A consequential missing condition becomes enabling work before deployment.

Economics and next decision

Estimate benefits and total costs as ranges, with adoption and timing. Separate released capacity from cash savings. Choose the next measurement, enabling task or pilot and state what could change the decision.

Questions executives ask

Do I need an AI idea before starting?: No. Start with a process, its outcome and an observed problem. The map helps you identify and compare candidate interventions.
Can the recommendation be a process fix rather than AI?: Yes. Process repair, integration and rules are legitimate alternatives. AI is useful when its capability addresses the constraint and performs reliably enough in representative tests.
Are recommendations or ROI validated predictions?: No. Decisions depend on your inputs, disclosed assumptions and delivery conditions. Replace estimates with operational evidence. The tool does not provide peer benchmarks.
Do released hours count as cash savings?: Released hours are capacity. Cash savings need an identified cost reduction, avoided hiring or another credible realization path. Review and rework must be included in net time saved.
Can I compare several processes?: Yes. Keep the outcomes, evidence and assumptions visible. Identify shared enabling work and avoid counting the same benefits in multiple opportunities.

Canonical URL: https://juanbeltran.ch/tools/ai-opportunity-map