Sometime back, in a fast-growing company, an important launch was getting delayed. Marketing felt R&D was taking too long. R&D pointed to changes in the brief. Supply Chain said some decisions affecting manufacturing were coming too late. Everyone had a reasonable explanation.

One person on the project started noticing where things were actually getting stuck. She saw that some decisions had been reopened, that changes in one area were creating delays elsewhere, and that different teams were working off slightly different assumptions. Over time, she became the person people turned to for the history, context, and connections between the moving parts.

None of this was formally her job. But she helped the functions work better together.

This is known as glue work: the invisible effort that helps people and teams connect, share context, build trust and coordinate in service of a common goal. It includes tasks like:

• Helping one function understand the other’s perspective
• Adding context and clarity to discussions
• Introducing two people who should be talking
• Helping new hires find their feet
• Following through when things fall between the cracks
• Nudging decisions forward

Glue work is rarely found in formal job descriptions – but without it, the efforts of individual employees can fail to come together into a coherent whole that makes sense for the organisation.

As AI accelerates individual output, what happens to glue work? Can some forms of it be eliminated or supported by AI? Will others become even more valuable because they depend on human judgement and relationships?

This week, let’s explore the evolving role of glue work as AI becomes more embedded in the workplace.

The current conversation about AI is focused on solo productivity. Can it help an analyst build a cleaner model, an engineer ship more code, a marketer generate better ideas? These are reasonable questions, but they look at productivity one person at a time, overlooking the vital work that happens between people.

What disappears when you ask AI first

Take something as small as asking a question. Earlier, if you needed to know something, you’d probably turn around and ask your colleague. Along with producing an answer, that interaction would enable a few other things. Both of you would learn what the other person was working on. Your colleague might introduce you to someone else familiar with the topic. A small amount of familiarity and trust would accumulate.

Today, if you have a question, you can just ask AI! It’s probably faster and takes less effort – that’s a real gain. But the interaction you no longer need was doing more than transferring information. It was building links, creating context and strengthening the team dynamic. Those are the advantages we miss out on in an increasingly individualised workplace.

Recent research on knowledge-intensive teams explores this dilemma. As people turn to AI instead of colleagues for information, advice and feedback, solo work may become more efficient while the collaboration network weakens. This means organisations may need to become more deliberate about which interactions still matter.

The glue AI is good at

There is some glue work that most of us would happily hand over to technology. AI can help hugely with tasks like creating meeting notes, sending reminders, consolidating updates, surfacing old decisions and preserving organisational memory. This allows people to spend less time relaying messages and acting as information conduits.

A field study at P&G found that individuals working with AI performed at a similar level to teams working without it. Plus, the use of AI helped bridge functional siloes and produce more balanced solutions. These findings suggest that AI could potentially assist with more sophisticated collaborative tasks. AI may not simply replace glue work; in some situations, it may become part of the glue itself.

The glue that remains human

At the same time, some glue work becomes more valuable because it depends heavily on judgement, relationships and a reading of context.

This includes:

• Noticing the unspoken: picking up on the tense undercurrent in a room and surfacing the concern that needs to be addressed.
• Unblocking decisions: asking the right question to move things forward when a discussion gets stuck.
• Fostering alignment: stepping back halfway through a project to check whether everyone is still solving the same problem.
• Connecting people: knowing that Peter in finance dealt with exactly the same issue as you last year and making the introduction.
• Translating across functions: explaining to Marketing that Technology isn’t saying no; it is worried about reliability at scale.
• Bringing context: sharing a piece of history or explaining a relationship that changes how a decision should be viewed.
• Sharing unwritten norms: helping a new colleague understand how things actually work versus how the handbook says they work.
• Building trust: repairing team relationships after conflict and shoring up enough trust for healthy disagreement.

These are non-glamorous, deeply human forms of glue. Where the work is documentation, synthesis, retrieval or routine follow-up, AI can take the lead. But where it involves trust, sensing, judgement, interpretation or a subtle reading of context and cues, the human contribution remains crucial.

Concentrated in a few hands

In many companies, a handful of people carry out the majority of glue work. They get pulled into every project review, every email thread, every cross-functional disagreement. They know all the ins and outs of the organisation and have a direct line to everyone.

Eventually, they become both indispensable and overwhelmed, with everyone relying on them for work that remains outside of their job description. Software engineer Tanya Reilly, who popularised the term ‘glue work’ through her viral essay a few years ago, highlights an uncomfortable truth: glue work is essential yet largely unacknowledged. An employee can become known as the one who holds everything together without receiving any formal recognition for their contribution.

This raises interesting questions for leaders. Can we make valuable glue more visible? Can we spread it more broadly? And can we outsource some basic coordination to AI to reduce the pressure on our human team members?

What leaders can do

Here are six recommendations to help you rethink how glue work gets done in your team and organisation:

1. Notice who enables others’ success.

Most talent conversations revolve around individual output. It’s also worth asking who consistently helps other people solve problems, think better and work across boundaries. These employees may not be the most obvious high performers, but keep in mind that their contribution may be spread across several people’s outputs.

2. Put it on the record.

Connecting colleagues, mentoring new hires and nudging forward decisions shouldn’t be treated like extracurricular work, done by whoever happens to be the most obliging person around. These activities should be part of the performance conversation and play a role in decisions around incentives and career advancement.

3. Watch for overdependence.

If everyone repeatedly turns to the same people for information or cross-functional solutions, this may point to a gap in how the organisation is wired. Consider broadening access to knowledge and creating more direct links between departments.

4. Hand the busywork to AI.

Use AI for things like meeting notes, status updates, basic follow-up and information retrieval. This frees up human time for glue work that demands subtler skills like persuading people, navigating ambiguity and resolving conflict.

5. Be intentional about adding glue.

Not every situation calls for a trust-builder or context-keeper. Consider where human collaboration and connection are genuinely important. In many cases, AI enables people to work well alone – there is no reason to manufacture glue work where none is required.

6. Protect interactions that build understanding.

Hold space for conversations where people learn how other teams think, uncover the history behind decisions and build the familiarity that makes future collaboration easier. For example, hold cross-functional reviews for complex projects, pair new hires with experienced buddies, and make time for those tough but ultimately rewarding conversations with team members.

As AI makes people more capable of operating independently, glue work will become more differentiated. Some coordination will disappear and some will become easier to manage with technology. The remaining glue will become vital to holding organisations together, proving most valuable in areas that call for human judgement, nuance and trust.

Who does the glue work on your team – and are their contributions recognised?