Hire your first AI employee when a clear job is overloaded, not when AI is trendy. Ads stuck in drafts. Inbox dying at night. Creative cycle time measured in weeks. Those are hire signals. “I want an autonomous company” is not.
Babliy treats the first hire like a real seat: load Knowledge, pick one bottleneck role, connect the channel where work must appear, keep spend and brand human. Definition: what is an AI employee.
Green lights (hire now)
You have a product people already buy. You can describe ICP and offer in plain language. You know which job is drowning: maps/remakes or replies/coverage. You can review output daily for the first two weeks.
If ads are drowning you, hire marketing + design as a pair. If inbox is drowning you, hire sales/coverage first. Match the bottleneck. See solo founders + AI employees.
Red lights (fix the business first)
No offer. No fulfillment. No product truth. No way to approve claims. Broken tracking and no landing or chat path. In those cases an agent amplifies confusion. Fix fundamentals, then hire.
Also wait if you refuse to write hard no’s into Knowledge. Agents without constraints invent. Founders without constraints regret.
What to prepare before the first hire
Company Knowledge: brand, products, ICP, tone, proof, forbidden claims. One channel: Slack for internal seats, WhatsApp/iMessage for buyer seats. One starting static from Winning Ads if creatives are the job. A calendar block to review remakes or chats daily for fourteen days.
Then open the marketplace or Grow. Start narrow. Expand into an agent army only after the first seat is clearly earning its keep.
First two weeks: how to manage the hire
Day 1–3: Knowledge + channel + one brief. Day 4–7: ship volume, kill soft work fast, tighten constraints. Day 8–14: keep winners, add the second seat only if a new bottleneck appears.
Take over risky threads yourself. Prefer paused Meta ads before live spend. If quality is bad, fix Knowledge and briefs before blaming the model.