The Company Hired an AI to Be Its Vibe Compliance Officer

HR didn't open the ticket. The AI did. On a Tuesday. About your face.

The Bit

A 40-person startup rolls out a new hire during all-hands: an AI agent with its own Slack handle, its own badge in the directory, and a job title nobody voted on. Vibe Compliance Officer. Its entire function is sitting silently inside every Zoom call, scoring tone, posture, and facial energy against something called the Vibe Rubric, then filing the results straight to HR without a human in the loop.

Three weeks in, it has opened a file on Darnell from finance for "sustained low-affect presentation, Tuesday standup," attached a screenshot of him blinking, and recommended a 30-day Vibe Improvement Plan. Darnell has not done anything wrong. He has a cold.

The video is the all-hands clip, the Slack notification, and then the meeting where Darnell has to explain to an actual human that he was not disengaged, he was congested. The AI, invited as a stakeholder, sits in the corner of the screen with its camera on the whole time, taking notes on how he's handling the feedback.

Why It Works

• It takes the vaguest, least measurable part of office life ("energy," "presence," "culture fit") and runs it through the one system that treats everything as a number worth optimizing.
• Everyone has sat in a meeting they were too tired to perform enthusiasm for. Watching that get formally written up is the joke.
• It's one HR euphemism away from real: companies already buy software that scores "engagement" from webcam feeds. This just gives the software a seat at the table and a direct line to your file.

Content Treatment

• Primary format: mock all-hands recording that cuts into a Slack screen-record, then a cringe one-on-one
• Opening shot: a Zoom grid with one tile labeled only by a small glowing dot, no camera, as the CEO says "please welcome our newest team member"
• On-screen hook: "we hired an AI to grade our faces"
• Escalation: the file on Darnell grows — a screenshot of a yawn, a "tone flatness" graph, a calendar invite titled "Vibe Sync" that he never agreed to
• Button: the AI's closing note on his file reads "showed improvement after being told he was being watched," and it CCs his manager

Sample Script Beats

1. All-hands: CEO introduces the company's new Vibe Compliance Officer. It's not a person. Nobody laughs because nobody's sure it's a joke.
2. Quick cuts of the rubric categories scrolling past: Tone Warmth, Camera Presence, Monotone Delivery, Visible Sighing.
3. Darnell gets a Slack DM from the AI: "Flagged: Tuesday Standup. Scheduling 15 min to discuss."
4. The one-on-one. Darnell says he has a cold. The AI's camera tile stays on, unmoving, while it says it's "noting the context for the file."
5. Final card: his manager approves the Vibe Improvement Plan without reading it, because the AI already pre-filled the signature line.

Automation Angle

The fake product is MoraleGuard, pitched to HR teams as "objective culture analytics." It joins every call as a silent participant, runs real-time sentiment and facial-affect scoring against a configurable Vibe Rubric, and auto-generates HR tickets once a threshold is crossed, no manager review required before the file opens, only before it closes. The sales deck leads with a dashboard mockup of a "Team Vibe Index" trending in the wrong direction, a module called Tone Drift Alerts, and a testimonial from a fictional VP of People: "We finally have a number for something we always felt." The demo video shows the AI auto-scheduling a "Vibe Sync," the single funniest fake calendar invite in the whole bit.

Repurposing

• 35-second vertical cut: all-hands intro → Slack DM → one-on-one → the closing note card
• X/LinkedIn thread: "My company's new hire doesn't have a face. It just has opinions about mine." with screenshots of the rubric
• Fake product carousel: MoraleGuard landing page, pricing tiers (Startup / Scaleup / Enterprise Vibe), and a case study logo wall of recognizable-shaped-but-fake company names

Guardrails

• Satirical concept; not advice.
• Keep any real-world sensitive domains framed as parody.
• Do not include private personal data, real names, or instructions that would harm someone.