playbook-new-in-role

Use the Prospeo search-person endpoint with the person_time_in_current_role filter set to a month range like {min: 0, max: 3} for a 90-day window. The filter is applied at the source, so returned rows already carry the role start date and job history needed for personalization.

The default is 90 days, expressed as a 0 to 3 month filter. If volume is too thin, widen the title family first, then headcount or geography, then the window to 6 months. Nine months is the ceiling because the just-started premise stops being credible beyond that.

The person_job_title filter is a loose contains match that can hit a past role, so results include people whose current title is unrelated. Gate in code on the current entry of the job_history array, and include abbreviations like COO in the gate list.

Empty output on every row is caused by finish_reason=length, where default reasoning effort burns the full token budget on reasoning. Set reasoning_effort to minimal and retry on length rather than recording it as an abstain.

Use the Apify-based robust pass only for high-value clients or verticals where the database is thin, since it costs roughly $3.20 per 1,000 profiles plus hydration. Its tenure filter is coarser and LinkedIn keeps stale results, so a cleanup pass on every run is mandatory.

Yes. About 1 in 10 lines can be a confident but wrong statement because the source database asserted a title the company's own site contradicts. The only control is an optional cross-check against the company's leadership page; structural QC cannot catch it.