The report is writable by AI, but the underlying data it needs is largely off-limits.

The structure is repeatable — count roles, extract skills, compare verticals, write report — but the specific verticals, time windows, and strategic framing shift each engagement, requiring fresh judgment each time.

The output format is reasonably specified (1,800 words, four metrics, three verticals, confidence levels), but 'most attractive' and 'most likely to need our services' are subjective calls that depend on the consultancy's own capabilities and positioning, which the agent doesn't know.

LinkedIn's API does not expose job-posting counts, salary data, or hiring velocity to third parties at this granularity. Scraping LinkedIn at scale violates its ToS and is legally contested. Proxy sources like BLS, Lightcast, or Burning Glass require paid subscriptions and are not natively available to a general-purpose agent.

This report will directly inform a major strategic expansion decision. If the agent fabricates or misrepresents market data — which is likely given data access constraints — the consultancy could invest in the wrong vertical, misallocate resources, or present flawed analysis to clients.

Recommending which vertical is 'most attractive' requires understanding the consultancy's existing relationships, team expertise, competitive positioning, and risk appetite — none of which an agent can infer from job-posting data alone.