AI is usually seen as a technology to automate tasks. But its impact goes further. It’s changing:

Let’s explore what we see changing in the market.

For decades, the workplace was designed around individual productivity: desks and private offices. COVID accelerated a reduction of personal space and an increase in amenities and collaborative spaces. CBRE's benchmarking shows individual "Me" space has dropped every year, from 56% of total space in 2021 to 35% in 2025. AI is now accelerating the move to more fluid working experiences, transitioning between deep focus work, group decision collaboration and recovery.

With AI, one person can produce what used to take a team. They draft, analyse and summarise with AI assistants. But deciding stays collective. Choosing between ten generated options, aligning a team behind one of them and judging whether it worked still need people in the same place.

Cushman & Wakefield makes a similar case in its ten-year outlook on AI and offices. As drafting and analysis get easier, judgement, accountability, negotiation and complex decisions carry more relative weight. The spatial consequence follows: less emphasis on rows of desks for individual output, more value in spaces for alignment and high-stakes discussion.

The space now has to support a new working loop: focused work alone, then a session together to decide, then back again. Several times a day. As Carlos Almansa argued earlier this year, AI shifts the question from where people work to what they need to do next.

Flexspace Observatory, our industry data project at Nexudus, tracks ten years of coworking bookings. The first pattern it shows is that spaces are being split into more bookable resources, offering more flexible experiences. This shift started with COVID. AI did not create it, but it is accelerating it.

In the largest coworking spaces, those with more than 250 members, the mean number of members per bookable resource was around 39 in January 2016. By mid-2026, it was around 23. Over the same period, members per space grew by 53%, while the number of bookable resources per space almost tripled. Operators divided the floor into many more bookable pieces.

In the US market, for example, small meeting rooms lead: over the last ten years, rooms for one to three people account for 23% of all bookings, rooms for four to five people for 18%, and rooms for six to seven people for 16%. The UK and other markets show a similar pattern. In the last six months, rooms for four to five people increased their share of total bookings by four percentage points. 

At the other end, in the global market, the large room is losing ground. Conference rooms dropped from around 20% of all bookings in 2016 to about 7% by mid-2026. Over the same period, booths rose from roughly 1% to 4%, and hot desks from around 2% to about 12%.

Bookings are also getting longer. The average booking rose from around 170 minutes after April 2020 to more than 180 minutes by mid-2026. Longer bookings could reflect more collaborative working sessions.

Most coworking floors were designed around the fixed desk as the main unit of work. The new type of work created by AI needs something else. It needs small spaces where one person can focus and talk to an AI assistant without the whole floor listening, flexible collaboration areas and small rooms where a few people can evaluate what the AI produced, and open spaces to relax and connect.

When the floor does not offer those spaces, members find workarounds. Small groups that cannot find small rooms occupy large ones, and a room for eight blocked by two people becomes a familiar sight. When they cannot find any room at all, they make do. In Gensler’s Global Workplace Survey 2026, 60% of workers hold meetings at their desks at least sometimes because no meeting room is available, and 43% cancel meetings for the same reason.

So the problem is not a lack of space. It is space in the wrong sizes. The layout is fixed, but the way it is organised and sold does not have to be. Fixing it does not mean knocking down walls. It means letting booking data decide what the floor offers.

Historical booking data is what allows coworking spaces to evolve their inventory and adapt it to current demand.

Getting the right mix of spaces is the first step. The next one is making it easy for members to move between them. In a recent article, Carlos points to where this goes next. A member moving through the loop does not arrive wanting to book a resource. They arrive wanting an outcome: "I need quiet." "I need my team together." "I need this client visit to feel smooth." That is why the next wave of workplace AI is not only automation. It is interpretation: understanding that intent and responding with the right mix of space, access and services.

For that to work, the space has to be built for people who work as they move through it. A member who goes from a focus booth to a team room, and then brings in a guest, should not have to make separate bookings for rooms, services, access and parking. They should be able to say what they need once, "I have a client at 12 and I need an hour with my team before that", and let the space, helped by AI, handle the rest: the room, access, the guest and parking. One intent should lead to one experience.

The operators who win this shift will understand behaviour, not just inventory. They will design around how people actually move through the day: focus, collaboration, privacy, recovery and community. For the member, the result is an experience that feels fluid, not fragmented.

For operators, AI that automates gives time back: fewer helpdesk questions, fewer emails about rooms, tours qualified automatically. But working with AI is intense. Workers, both operators and members, create, review and evaluate AI outputs, then move straight on to the next prompt, which adds mental fatigue. According to Quantum Workplace data shared by Steelcase, frequent AI users report more burnout than infrequent ones: 45% vs. 38%.

That is why the new working loop needs a third step. Make alone, decide together, then recover. Workers are already asking for it: physical and mental wellness amenities top Gensler’s list of hopes for the future workplace, at 46%. As AI use grows, that demand will only increase.

Operators don’t need a new layout. They need a new mix. To get it right, they need to know how members actually use the space they already have, and which resources they need most. That is what booking data shows, and where Nexudus can help.

That's it for today. See you in two weeks! 🙂