Onboarding is the process that brings a new person into a role, from their first day until they can work on their own, covering knowledge, tools, relationships and supervised practice. In teams that work with AI agents, onboarding ends only when the person can decide the agent's held actions in their area without someone checking each one.
Why does onboarding matter in a team with agents?
The skills a new team member needs have shifted. Less of the job is doing the work by hand; more is checking, deciding and correcting what an agent drafts. A newcomer who is onboarded only to the old process will either approve too readily or hold everything.
Onboarding is also where habits form. People who learn from the first week to read the verdict line, check provenance and record their reasons keep doing it. People who learn shortcuts keep those too.
What should onboarding include for agent work?
A practical onboarding plan for an agent reviewer covers:
- Week one: context. Which agents the team works with, at what depth, and the red lines.
- Week one: workspace. A guided tour of the queue, the record and the stop control.
- Weeks one to two: concepts and labs. Depth, the five bounds, the four verdicts, provenance, practised on seeded items.
- Weeks two to four: supervised live review. Each decision checked by a qualified reviewer.
- End: assessment. A test on real or realistic held items, then approval rights within the person's authority.
Checklists make this repeatable. See the reviewer onboarding checklist for onboarding teams and the equivalent checklist for finance teams.
What does onboarding look like in practice?
Here the word has two meanings, and in some teams both apply at once. Take a bank's customer onboarding team, where an agent gathers KYC documents and drafts account-opening decisions for a person to approve. A new team member is being onboarded to the job of onboarding customers. Her plan includes the bank's general induction, then labs on held account-opening decisions, then two weeks of checked review.
Part of what she learns is what makes human review meaningful, rather than a rubber stamp, where GDPR Article 22 applies to decisions based solely on automated processing; that is covered in GDPR Article 22 and training reviewers in KYC and onboarding. The team's own training is set out in training the onboarding team to review agent decisions.
How is onboarding different from induction?
Induction is the formal start: policies, systems access, introductions, often in the first days. Onboarding is the whole period until the person works independently, which for agent reviewers includes assessed practice and can take weeks. Induction informs; onboarding builds ability and ends with evidence of it.
Which related terms should you know?
- Shadowing: watching a qualified reviewer before deciding alone.
- Incident drill: worth including late in onboarding, as in the onboarding team's incident drill.
- Customer onboarding: bringing a new customer into a service, a common job for agents.
For the full approach, see training teams for AI agents.
Where the Academy fits
Pryme Intelligence Academy fits onboarding in two steps. The Foundations track, about four hours, gives a newcomer the vocabulary in the first week. Certified Pryme Operator, with its two-week practicum of real review decisions signed off by a Pryme assessor, marks the end of onboarding for someone who will run a live workspace daily.
Questions people ask
When does onboarding end?
When the person passes an assessment on the decisions their role requires and is given approval rights. A fixed date is a poor proxy.
Should experienced hires go through the full onboarding?
They can move faster through concepts, but should not skip supervised review on this team's agents. Experience elsewhere does not show how they read this workspace.