01

Start with a job, not a model

An AI employee starts with a job your team can describe and supervise. It might qualify new leads, answer support questions, handle appointment requests, collect intake details or keep a work queue moving.

The AI model matters, but it’s only one part. The setup around it decides whether the employee is useful, consistent and accountable.

02

The six parts of a working AI employee

Aivah treats an AI employee as one connected system. Each part answers a practical business question before you put the employee to work.

  • Job: what result is the employee responsible for?
  • Knowledge: which approved documents may it use?
  • Presence: should it work in text, by voice, on the phone or as a live character?
  • Apps: which actions and systems is it allowed to use?
  • Memory: what should it remember, and under which rules?
  • Management: who reviews conversations, results, exceptions and handovers?
TRY IT WITH YOUR TEAM

Try the job you would give Aivah.

Choose sales, support or onboarding. Each example shows its source brief, so you can judge the answer.

Choose your business demo
03

Why the channel changes the work

A job feels different on a landing page, a phone call, a meeting or a live character on your site. It changes again when your own team messages it in Slack, or you do on WhatsApp. Done well, each channel keeps the same knowledge and limits while the conversation fits the place.

That’s why publishing isn’t just the last technical step. It’s part of designing the employee. Answer length, identity, handovers, consent and context all need choices for each channel.

04

Pilot one job you can measure

The safest start is one contained job with a result you can see. Decide what success looks like, what the employee may do, when a person takes over and what your team will review.

Once that works, the same employee can take on more knowledge, channels, apps and responsibility, without losing its limits.