AI can crunch the survey data and draft the briefs, but a consultant needs to own the story.

The structure is identical every quarter: same 30-item Likert survey, same segmentation dimensions, same deliverable format. This is a strong signal for automation — the agent can be built once and reused with minimal reconfiguration.

The quantitative outputs (lowest-scoring items by segment, peer benchmarks) have crisp success criteria. The narrative analysis is underspecified — 'unique engagement risks' and 'peer benchmarks' require judgment calls about what's meaningful versus noise, and what tone is appropriate per client relationship.

Survey response data must be piped in cleanly (CSV, database, or survey platform API), and a visualization library or BI tool must be accessible. If the agent is handed structured data and has Python/R or a BI connector, this is tractable — but data access and permissions across 28 clients is a real integration hurdle.

Misattributing a score to the wrong company, inverting a benchmark comparison, or generating a narrative that mischaracterizes a client's culture risk could damage the consulting firm's credibility and client relationships. These are external-facing deliverables with real reputational stakes.

Translating statistical patterns into actionable, client-specific engagement narratives requires knowing each client's history, sensitivities, and organizational context — none of which lives in the survey data. A consultant also needs to decide which findings are worth surfacing versus which are statistical artifacts.