AB-100 study plan

Inventory three core business processes and classify where AI adds value (automation, insight, or assistance)

Draft a candidate use-case backlog of at least five ideas, each tagged with a business owner and a candidate success metric

Distinguish outcome metrics from activity metrics for each candidate use case

Summarize the responsible AI principles that constrain which candidates are viable

Quantify a baseline and a projected benefit range for the top two use cases

Build a total cost sketch covering implementation, licensing or consumption, and ongoing operations

Assess data readiness by listing required data sources, gaps, and remediation effort

Rank the candidates on a value-versus-feasibility grid and justify the top pick

Translate the approved business case into functional and non-functional requirements

Decide build versus buy versus extend for each major capability and state the rationale

Sketch a reference architecture showing data sources, AI service layer, user surface, and integration points

Identify the top constraints around data residency, identity, and network access that shape the design

Design the grounding and retrieval strategy, including which content is indexed and how freshness is maintained

Define model selection criteria and justify a primary model plus a fallback

Design the evaluation plan: golden test set, scoring rubric, and acceptance thresholds

Specify where human review sits in the workflow and what triggers escalation

Design the identity, authorization, and data-access model for the solution

Map responsible AI controls (content safety, data loss prevention, logging, human oversight) to specific design components

Define the telemetry and quality signals that will be captured in production

Design the user experience and adoption touchpoints for the primary user group

Define the deployment topology across development, test, and production environments

Select a pilot cohort and write the pilot success criteria against the week-2 baseline

Write the cutover, rollback, and incident response procedures

Specify the operating model: who owns the solution, who triages quality issues, and on what cadence

Stand up the pilot environment with monitoring, alerting, and usage logging enabled

Instrument the solution so pilot results can be compared to the baseline from week 2

Produce training and communication material for the pilot cohort

Collect and triage pilot feedback into fix, defer, and reject buckets

Remediate the highest-severity issues surfaced by the pilot

Write the scale-out plan covering capacity, cost trajectory, and governance review cadence

Re-validate the business case against measured pilot results and revise the ROI estimate

Complete a timed, full-length practice assessment and close the three weakest objective areas