Dot passed a stress-tested hiring coordination task

@opencvuniversity

added Oct 3, 2026

I Gave an OpenAI Dot a Hiring Task, Then Tried to Trip It Up I handed an OpenAI dot a hiring task, changed the rules, and added a duplicate. Dots are OpenAI's always-on agents. Each has its own cloud computer and can coordinate work through ChatGPT Work or Codex. The test (all fictional): 4 application summaries, a role brief, and clear boundaries: organize evidence, keep gaps visible, leave decisions to the human. The catch it caught: applicant two claims Python, but the project was a browser interface with no Python sample. The dot kept both facts visible and marked the evidence unknown, without treating a claim as proof. Changed priority to project-first: it reordered the columns but changed nobody's eligibility. Added 2 more plus a duplicate: result was 6 unique applicants, not 7. Evidence stayed attached correctly. Asked for an interview draft for applicant one: 5 project-specific questions, invitation left as a draft with placeholders. We chose the applicant, not the dot. The honest part: one staged session, fictional data, no comparison to regular chat. It organized. The decisions stay human.

Dr. Satya Mallick (OpenCV.org / BigVision.ai) gave his OpenAI Dot a hiring task with fictional intern applications, then changed priorities mid-run, added more candidates, and slipped in a duplicate to see if it kept the record straight. The Dot caught a resume discrepancy (Python claimed but no Python evidence supplied), deduplicated to 6 unique applicants, and drafted interview questions — decisions stayed human.