Learning Library · Reference map

AI Business Solutions Map

A compact reference for the kinds of business problems AI consultancies work on today. It is a map for breadth, not a checklist of things I need to build.

Reusable pattern

Find→Extract→Classify→Synthesize→Recommend→Generate→Decide→Act→Monitor

Not every solution uses every step. The useful question is which parts need AI judgment, which should be deterministic, and where a person or trusted system retains authority.

Solution families

Where the work tends to show up

Knowledge & research

Enterprise search, research assistants, project knowledge, competitive intelligence.

Documents & content

Extraction, classification, contract review, claims processing, RFPs.

Workflow & agents

Routing, approvals, multi-step operations, and controlled system actions.

Customer & revenue

Support, sales, marketing, personalization, and customer-facing assistance.

Back-office operations

Finance, procurement, HR, legal, compliance, and risk work.

Technology

IT, cybersecurity, software development, and analytics.

Physical operations

Manufacturing, field service, logistics, and supply chain.

Industry-specific applications

Insurance, healthcare, retail, government/public sector, and R&D.

Product questions that repeat

The domain changes. The PM questions often do not.

ReliabilityWhat must be right, and what kinds of failure matter most?

Context & dataWhat information is authoritative, current, and actually needed?

Human approvalWhere can AI assist, and where must a person decide?

PermissionsWhat can the system see or do, under whose identity, and with what limits?

Failure handlingWhat happens when the model, data, tool, or workflow is wrong or unavailable?

Measurement & ROIWhat user or business outcome should improve enough to justify the complexity?

MonitoringHow will we detect silent quality problems after release?

RolloutWhat should start narrow, and what evidence would justify more scope or autonomy?

Contexts practiced so far

A breadth marker, not a completion checklist.

I can add to this as exercises use genuinely different business contexts.

Project knowledge · StateCustomer support · MeridianLegal / compliance · Legal AISoftware / product development · State testing, evals, tracing, architecture