Onboarding was always partly a bluff. A new joiner spent six weeks appearing competent while quietly working out who to ask, which system held the real numbers, and which documented process everybody ignores. The documentation helped a little. The person sitting next to them helped enormously. Assistants change the first part of that and leave the second untouched, which produces a new joiner who reaches apparent competence much faster than actual competence — and an organisation that stops noticing the difference.
A new joiner with a good assistant can produce correct work in week one without understanding any of it. That is a genuine gain and a genuine risk, and they arrive together
Here is how to redesign onboarding for that reality.
What the assistant genuinely replaces
The lookup layer, almost entirely. Where the policy lives, how the expense process works, what the acronym means, which form to use. This was a large share of the first month and a large share of the interruptions the existing team absorbed. The first draft of routine work, which lets a new joiner contribute output before they could previously contribute anything. And the question they were embarrassed to ask, which is a real and underrated benefit — people ask a machine things they would not ask a colleague in week two.
What it does not replace
Judgement about what is normal here. Whether this supplier is usually late, whether this customer always disputes, whether this number being high is interesting or routine. That is pattern knowledge accumulated by exposure and it cannot be retrieved. The informal map of influence and reliability, which is most of what makes someone effective in an organisation and is written down nowhere. And calibration — knowing when their own output is wrong. A new joiner cannot evaluate an assistant's answer in a domain they do not yet understand, which is precisely the situation they are in.
The redesign
Stop spending the first week on information transfer; the assistant does it better and on demand. Spend it on context that is not in any document: how decisions actually get made, who to trust on what, what has gone wrong before. Make the assistant part of the induction explicitly — teach people what it is good at and where it is confidently wrong in your environment, because they will otherwise calibrate by accident. And change what you check. Reviewing a new joiner's output tells you less than it used to, because the output is assisted. Ask them to explain the reasoning instead, which is the only way to see whether understanding is forming underneath the competence.
| Onboarding job | Design emphasis |
|---|---|
| Find a reference | Teach the assistant's source limits and failure modes. |
| Draft routine work | Review the output and the new joiner's reasoning. |
| Handle an exception | Use a named buddy and deliberate exposure to judgement cases. |
| Check understanding | Ask why and what would change the answer; keep reviewing competence. |
Qualitative summary of this article's source text, not a measured outcome or performance estimate.
Practical Guidance for Onboarding Modernization
- Cut information transfer from the induction week.
- Front-load the unwritten context — people, history, norms.
- Teach the assistant's failure modes explicitly.
- Review reasoning, not output, in the first months.
- Keep a named human buddy; the assistant does not replace one.
- Use the questions asked as a documentation backlog.
- Design deliberate exposure to the judgement cases.
- Re-check competence at six months, not six weeks.
The Regional Angle
The first regional consideration is that onboarding in the Gulf carries an administrative burden that no assistant touches. Visa processing, medical checks, Emirates ID or iqama issuance, bank account opening and labour contract registration consume weeks of a new joiner's first months and are largely outside the employer's control. Organisations focusing their onboarding redesign on knowledge transfer while leaving the administrative path unmanaged are optimising the part that was already improving and ignoring the part that actually determines how the first month feels. The second concerns the scale and pace of hiring in regional growth markets, where teams frequently double within a year and the informal transfer model simply fails. When half the team joined in the last twelve months, there is no experienced neighbour to absorb the questions and no accumulated sense of what is normal to transmit. That is the environment where assistant-supported onboarding delivers the most and where the missing judgement layer is most dangerous, because nobody in the room has it either. The third is about language and the confidence gap it produces. A new joiner working in a second language will ask an assistant questions they would not raise in an open-plan office, and for a significant share of the regional workforce that is a material improvement in how quickly they become useful. It also means a supervisor may see fluent, well-formed output from someone who has not yet been able to express uncertainty — which makes the shift from reviewing output to discussing reasoning more important here than elsewhere.
The objection worth taking seriously
The strongest objection is that this describes a training problem and then declines to solve it. If new joiners can produce acceptable work without understanding it, the understanding may simply never arrive — the difficulty of the first months was not an inefficiency, it was the mechanism by which people learned. Removing the struggle removes the learning, and an organisation optimising for time-to-productivity is trading a measurable short-term gain for a workforce that is permanently dependent on tools it cannot evaluate. That mechanism is real and the concern about dependence is well founded. The response is that the struggle was a byproduct, not a design, and byproducts can be replaced deliberately. Most of what a new joiner previously suffered through was searching for information, which taught nothing except where things were filed. The part that did teach — exposure to cases requiring judgement, being wrong and corrected, seeing how decisions were made — is still available and is now affordable in week two rather than month four. What the objection correctly identifies is that this does not happen automatically: an organisation that removes the friction and adds nothing will get exactly the outcome described. The answer is to design the exposure, not to preserve the inefficiency that used to produce it accidentally.
Common Questions
Should new joiners be restricted from assistants initially?
No, and the attempt is unenforceable. Teach them where it fails in your environment instead, which is more useful than any restriction.
How do we tell whether someone actually understands?
Ask them to explain why, and to identify what would change the answer. Both are difficult to fake and neither is improved by an assistant.
Does this shorten onboarding?
It shortens time to useful output substantially and does not shorten time to real competence at all. Planning against the first number is the mistake.
What should we expect over the next twelve months?
Expect induction programmes to shrink and then be partly rebuilt. Expect the six-month competence gap to be discussed openly once the first cohort reaches it. Expect assistant-usage guidance to become standard induction content. And expect the buddy role to prove more durable than anyone predicted.
Onboarding Modernization — we rebuild the first month around the context nobody wrote down, because the rest is already handled.