Skip to content
Interface language: English
Imperium / Signal

A team of strong specialists: what capability could still be missing?

Look at a team's tasks, access to expertise and how contributions connect. Find the gap between an AI-assisted draft and an approved result.

Article language: English

Concept illustration: a woman rests her hands on the outer two of three round discs mounted on an axis in a navy and red interior.
In this article

Four people present excellent drafts prepared with AI. Each can produce several versions of a text and revise them quickly. Yet when a product description needs approval, everyone waits for someone to compare it with the manufacturer's current material. The team has no shortage of ideas. It has not agreed who will do a particular piece of work.

Imagine a company assembling a small team to produce an upholstery fabric catalogue. This is a hypothetical example. The team members have editorial experience and use AI tools confidently. But the company needs more than drafts: its descriptions must be based on the correct versions of the data. In the first allocation of work, everyone receives several products. No one is clearly assigned to resolve differences between a draft and its source.

The leader might conclude that another strong hire is needed. First, it is worth identifying which capability the team as a whole lacks and when it must be available. Another person producing drafts will increase their number, but may not close the approval gap. Discussing team composition through the work that needs doing is more useful than simply adding up impressive CVs.

A list of skills does not describe the whole job

Map the result the company actually needs. For the catalogue, that means describing the right product, connecting the description to a current source, resolving uncertainties and handing the material over for publication. Different people may do each part. What matters is that each has the skill, access and time needed at the right moment.

Do not start by multiplying job titles. The title "editor" can cover very different experience. One person is excellent at explaining a material's use, another recognises manufacturer markings, and a third organises information on a product page. Each may also do more than one task. Identify actual capability rather than assuming it from a job title or familiarity with a tool.

In research published by Woolley and colleagues in 2010, 699 people worked in small groups on different tasks. The authors distinguished individual results from a common statistical factor describing group performance and also examined how people interacted. The research did not concern making catalogues with AI. It does, however, give reason to be cautious about treating a group's result as a simple reading of its members' individual results.

A software skills test or an assessment of one draft can show that a catalogue team has a needed skill. It will not yet show whether the team can bring the shared material to completion. That decision also needs a description of dependencies: whose decision lets the next person act, and who can provide it within the agreed scope.

A job interview (in Polish) can look for evidence of the missing work. First, though, the company has to know what it is looking for. The general requirement "another excellent AI person" does not establish whether the team needs someone for data, subject-matter judgement or organising the handover. It might not need a hire at all, only a different allocation of the work already available.

Choose one handoff between activities and name the contribution it needs. For the first catalogue description, someone must confirm which product the source concerns. For the second, someone must decide whether new data replaces the previous version. These are different questions. The team may have one person able to answer both, but it should know how to direct the right case to them.

One draft calls a fabric "easy to clean". The editor sees this as a useful explanation for the buyer, while the person who knows the materials asks what supports the statement. The source concerns another property and does not establish that claim. Two contributions are needed: recognising the boundary of the available information and preparing an honest description for the reader. Replacing the writer with a better stylist will not supply the missing basis. Pointing out the problem alone will not produce a finished text either.

The team agrees that the subject-matter question remains open and that, for now, the description will use only confirmed data. The editor prepares a new version, the specialist checks the specified scope, and the publisher can see which material the approval covers. None has to do everyone else's work. They do need to understand what each person hands over and what the recipient should not infer beyond that contribution.

Knowledge must reach the place where it is needed

In a study of 69 software development teams published in 2000, Faraj and Sproull analysed expertise coordination: recognising where expertise resides, when it is needed and how to bring it into the work. In their cross-sectional analysis, it was associated with performance over and above the presence of expertise and administrative coordination. This is an association in a particular sample, not proof of a universal effect from changing team composition.

For the catalogue team, this raises an organisational question: is it clear when someone who knows the material data needs to contribute? That person may appear on the project list and still not take part at the required moment. If they receive the whole catalogue only just before publication, earlier drafts may already rely on unagreed assumptions. An expert's presence on the team does not by itself describe when their contribution is available.

Agree how a question is directed. The editor should be able to point to the source, the product description and the specific discrepancy. The specialist should not have to reconstruct the entire context from a message saying "please check everything". The person preparing the draft should know whether the answer resolves one item, a whole family or only the meaning of a particular entry.

Aeneia conceptual photography

In Kyle Lewis's 2003 research, a scale describing a team's system for using knowledge was tested in three samples: laboratory teams, MBA consulting teams and teams from technology companies. The abstract also points to the need to test the measure's limits further. It is a reminder that recognising a shared capability requires a way of assessing it, not merely calling it "teamwork".

You do not have to turn a small project into a psychometric study. You can follow the actual path of preparing one description and see whether the knowledge needed reaches the right person. Separate an observation of the work from a general statement that the team collaborates well. Only then can you choose a change that addresses the actual gap.

The same precision helps when asking for help in a team (in Polish). Finding the source, checking an interpretation and approving an exception each require a different contribution. A broad "can anyone help?" may always reach the person who replies fastest, even when someone else's expertise fits the case better.

Repeating a strength does not fill a missing function

In the first week, all the editors efficiently produce versions of the descriptions. At the end of the day, the leader sees a large number of finished files. When they ask about sources, it turns out that some people used the previous catalogue and others used new product sheets. The problem appears only when the shared result is compared. Individual efficiency was not enough to maintain one basis for the work.

That does not mean permanently assigning one person exclusively to checking. The team might work in pairs, bring in a specialist for defined questions or reduce the number of items being prepared in parallel. The choice depends on the knowledge and time available. Each option should still identify who does the necessary work, rather than leaving it to everyone's presumed diligence.

Check that person's workload. If one specialist must confirm data for every description while also producing a full batch of her own, the plan may be unworkable despite a correct list of skills. Choose the batch size, divide the tasks or provide other support. Writing a name next to an activity does not add hours to the week.

That work can be invisible in a list of completed texts. Recognising contribution to preparation and checking (in Polish) helps show what made other people's drafts usable. If the company assesses only the number of descriptions signed off under a person's name, the person resolving source questions may appear less productive precisely because they are sustaining the work of the whole team.

Do not promise interchangeability merely because two people can access the same files. The second person must understand the question and be able to act within the agreed scope. Sometimes a prepared example and a shared walkthrough will suffice. Sometimes the current team does not yet have the required knowledge. Distinguish these situations before planning complete independence.

A leader's role here requires a different view from quickly correcting every description personally. In the move from expert to team leader (in Polish), it is necessary to decide what should be produced through other people's work. If every difficult question ends with the leader taking over the whole text, the company still does not know which capability is missing from the process.

Walk through a shared task before changing the team

Prepare a small batch with an ordinary product and a case where the materials disagree. Ask the team to take the whole route to a description ready for review. Do not limit the exercise to generating an appealing paragraph. You also want to see the choice of source and the point at which an editor directs a question they cannot settle alone.

Record who did each activity and what they needed from others. If two people checked the same passage, establish whether the second check had its own purpose. If no one compared the description with the material, record the gap without assuming bad intent. Check the authority of the person who spots a problem too. If they can only report it and no one decides before publication, the team has knowledge without an effective way to use it. Adding another checker will duplicate such reports until the company establishes who decides the next step. Perhaps everyone thought that was the publisher's job, while the publisher expected already approved content.

Choosing between proposals also needs an agreed method. When assessing a concept without the author's name (in Polish), some information about its origin can be separated from the first reading. For the catalogue team, however, the more important unknown may be consistency with the source. First establish what the check concerns; only then decide who should perform it and what information they need.

After the exercise, separate gaps in knowledge from gaps in access and time. If an editor can check the source but never received it, training alone will not solve the problem. If she received the material and still does not understand a marking, the right expertise or support is needed. If she can do the check but has a dozen other commitments at the same time, changing the instructions may not be enough either.

A company can bring a concrete account of how it uses AI to Aeneia Audit. The assessment will consider the roles of the tool, sources and people in the task, and the points where the result lacks an appropriate check. The company will make its own decision about team composition and workload. The material will help establish which capability needs to enter the process.

The next change might be small: involve the person who knows the data earlier, check selected items separately, or limit the batch to what the team can verify. Walk through a similar task again and see whether the gap has closed. A well-composed team must do more than produce files. It should carry the necessary work through to a result someone can use responsibly.

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