Skip to content
Interface language: English
Imperium / Signal

Onboarding an employee into company knowledge: tasks, sources and checking independence

Build onboarding around role-based tasks, current sources, help and evidence that a new employee recognises the limits of their authority.

Article language: English

A designer tests movable wall modules and benches on open terraces overlooking the water.
In this article

A new person gets access to a drive containing a folder called “Important”, a subfolder called “New” and a file named “final_final_v7_corrected”. That is a reasonable opening for a detective novel and a weaker start to independent work. Base knowledge onboarding on the role's tasks, current sources and access to help. Check whether the employee can find an answer and recognise the limits of their authority. A ticked list of documents read does not establish that.

A well-designed process helps the mentor too. Instead of repeatedly answering “which file is right?”, they can observe how the newcomer works and improve places where documentation does not lead to a decision. The aim is gradual independence with recognition of authority limits. It does not mean a duty to solve every unusual situation alone.

Choose tasks representing the real role

Start with several regular activities: preparing a draft offer, receiving an enquiry, finding a product requirement or handing over a project. For each, describe inputs, required sources and expected result. Also choose one exception requiring consultation. Onboarding then covers both the normal route and the moment when an employee should stop and seek a decision.

Ask an experienced person what they actually check before doing the task. Some rules may exist only in their memory. Record them in plain language and verify them using an example. Do not immediately rewrite a many-page manual. A short instruction used in a trial will quickly reveal a missing condition, definition or source. Expand it afterwards if useful.

If a task always returns to one expert, separately check whether another person can provide cover. The article on dependence on one expert (in Polish) discusses how to test that ability. Access to instructions and attendance at a demonstration do not yet prove a second person can recognise an exception and work within the agreed level of independence.

Organise sources using your company’s knowledge architecture. The guide to organising company knowledge (in Polish) explains how to structure sources. That step clarifies current versions and document owners. Onboarding checks whether a new person can use them for a specific task. A good library does not guarantee a good learning path, just as a large parts catalogue does not automatically teach someone to assemble a device.

Prepare a short source path instead of one folder containing everything

For each task identify the main source and supporting materials. Explain how to recognise the current version and who confirms its content. If older documents exist, label them clearly for someone unfamiliar with company history. A new employee should not select a version by file appearance or whichever link appears first in search results.

Add a role map: service owner, process mentor, person approving an exception and technical support. Explain what to bring to each. A list of names alone does not explain responsibilities. Information about cover is also needed. The process should work during an ordinary absence; otherwise the newcomer's ability to learn depends on one person's calendar.

Microsoft describes SharePoint employee-onboarding templates as a way to organise content, tasks and help for newcomers. Such a portal can be an entry point if materials fit the organisation. Microsoft: new employee onboarding site. Choosing a tool does not replace the tasks and independence criteria in this guide.

Teach use of sources by completing a task

Carry out the first trial together. The mentor demonstrates how to recognise the problem, find information and check its currency. The newcomer then performs a similar task with help. The next step is independently preparing a result for review. Record where a hint was needed and whether it concerned knowledge, access, navigation or an unclear instruction. When choosing the level of support and learning material, account for prior knowledge. The guide to choosing learning materials and examples (in Polish) discusses how to tailor support. A newcomer to the field and an experienced practitioner learning a new tool may need different explanations.

Do not assess speed alone. Someone who quickly selects the wrong price list is not more ready than someone who takes a moment to check the source. Agree a reasonable level of care for the role. If a task affects a customer, the employee should be able to identify the basis for an answer and recognise a case needing approval. That practical ability can be observed.

A trial card can include:

  • The task, inputs and scope the employee may perform alone.
  • Sources actually used to prepare the result.
  • How currency and suitability for the case were checked.
  • The completed result assessed against explicit criteria.
  • Hints required and their causes.
  • An exception correctly referred to the appropriate role.

When an error appears, return to where it arose. An employee may not know a term, but two documents may also conflict. Do not turn every issue into a recommendation to reread the instructions. Onboarding tests company knowledge too. A recurring question from several newcomers often points to a gap in material or how it is found.

Recognise access and responsibility limits

A new person should get the access needed for tasks at that stage. Access to every archive “just in case” makes navigation harder and may exceed the role's needs. Establish which materials are public, internal or require separate approval. Use properly prepared exercise data, especially where customer information or restricted past projects are involved.

If the process uses AI, explain what can be sent to the tool and what must be checked in its answer. Use your company's approved AI-use policy. If approved rules are not yet in place, establish them first with the people responsible. Do not assume a newcomer can reconstruct the rules from colleagues' behaviour. Team practice may differ from the approved process, and onboarding should lead to the correct rules.

Aeneia conceptual photography

For guidance on defining those rules, see the article on a company AI-use policy (in Polish).

Also agree the scope of independent action. Preparing a draft message does not necessarily include sending it. Gathering data does not mean permission to share it further. The distinction in responsible agent workflow principles helps with human work too: the task, data and approval point must be clear before an action with an external effect. As a related analogy, the agent workflow article (in Polish) discusses evidence of completion instead of relying only on a “done” message.

Hypothetical scenario: a new member of an offers team

Imagine a hypothetical company onboarding someone to prepare service offers. This is a working example without real employee data. The first task is to draft an offer for a standard case. The library contains two similar presentations. The newcomer chooses the older one because its filename is clearer and it appears first in search.

The mentor does not end the assessment by calling them careless. They check why an archived document looks like the current basis. The team improves the label and link in the onboarding path. In the next trial, the employee must identify the version and source for the service scope before writing a proposal. This teaches how to recognise the basis rather than memorising that “that particular file was wrong”.

The next case requests an unusual deadline. The newcomer recognises the exception and refers it to the service owner. That is a correct onboarding outcome even if the draft needs another decision. The company records the reason in the decision register so a similar situation can later become prepared training material without revealing customer data.

The project decision register guide (in Polish) describes how to record the reason for a decision and its context.

Test independence on a new case

After practice, repeat the trial with a different but comparable task. Do not copy the exact earlier scenario or you will test memory of the answer. Change one condition and see whether the employee chooses the right source and recognises the boundary. If they use AI, assess how they verify the answer. The guide to training a team to check AI answers (in Polish) discusses how to practise that skill.

Agree several readiness levels: work with help, an independent draft for review and independent completion within a defined scope. State the tasks each covers. Someone can be ready for one process while still learning another. A single broad “onboarded” label may be too imprecise for a manager to assign work sensibly and plan support.

Collect questions and material corrections in one place. Assign an owner to each gap and mark when the new version was checked. Not every comment needs a new document. Sometimes a name, link or example is enough. If the problem is an unclear business decision, that decision must be resolved. A larger library cannot replace a missing agreement.

Accept the process from the mentor's perspective too

Check that a mentor can prepare a trial without creating everything again. They need current tasks, assessment criteria and a place to record results. Assess support time and the load on approvers. If onboarding requires one expert's constant availability, arrange cover or simplify the first tasks. Otherwise the schedule depends on a resource that cannot be guaranteed.

The first tasks also quickly reveal the need for your own company glossary. The guide to creating a company glossary (in Polish) explains how to assign and maintain definitions. Record words different departments understand differently and illustrate their meaning with an example. A newcomer should not discover three definitions of “done” only during the first handover.

What to do with the newcomer's feedback

Collect comments close to the task. “The knowledge base is unclear” does not yet identify a fix. Ask what the employee searched for, which result they chose and why. This helps distinguish a problem of naming, structure or content. Do not expect the newcomer to design the solution; their observation is evidence of a user experience while learning the company.

Compare the comment with other people's behaviour. If the same mistake recurs across onboarding sessions, inspect the material and how it is found. If it concerns one unusual case, an example or boundary may be needed rather than a whole-library revision. The knowledge owner should decide the scope of the correction so suggestions do not produce conflicting instructions added without a shared review.

After correction, run a small trial. Ask someone who does not know the earlier answer to complete the task with the new instruction. Check whether they find the source, understand the condition and know when to consult. An author's acceptance of an edit alone does not prove a user can perform the task. The result may call for another small change rather than a broad rewrite.

Keep positive observations too. If a short example helps everyone recognise an exception correctly, use a similar format in later tasks. Do not automatically expand it to every conceivable situation. Onboarding material should guide the basic work and direct people to further sources when needed. Its effectiveness comes from use in a task, not document length.

Finally, compare the prepared plan with what actually happened. Which sources were needed, what was unnecessary and where was an example missing? Keep useful corrections for the next person. If onboarding includes AI tools, explore Aeneia Audit. Begin with a task and material that let you check how the result is used.

Aeneia Audit Describe a task being taken over by a newcomer and the AI tools and materials used. Discuss your needs and the possible scope of a review of AI use and information flows. Discuss the scope of an AI review

Aeneia / Aeneia editorial

Editorial article. Source links are included in the text.

About Aeneia

Brands we have worked with

  • L’Oréal
  • Garnier
  • Netflix
  • Carrefour
  • Warner Bros. Discovery
  • Biedronka
  • HBO Max
  • Wolt
  • Żabka
  • COSRX
  • Embassy of China in Poland
  • Mixa
Aeneia / Initium

From knowledge to a decision

Tell us what you want to put in order. We will agree on where to start.

Contact Aeneia

We partner with and use solutions from:

  • Google
  • OpenAI
  • Microsoft
  • Stripe
  • Anthropic
  • Linear