Can a Coworking Space Dream?

Today I read about an AI agent harness developed at Google that can retain memory across conversations, learn from previous interactions, and improve its understanding over time.

It made me think about a broader question:

Can a coworking space dream?

If it could, what would it dream about? Its members, the team, meetings, sales conversations, maintenance issues, payments, events, and the decisions made during the week?

Some theories in psychology suggest that dreams help us process experiences and consolidate memories. Anyone who remembers defragmenting a Windows 95 computer will understand the analogy: scattered information is reviewed, reorganized, and made easier to use.

What would that process look like inside a coworking business?

AI can already prepare task lists, review calendars, organize meeting notes, draft contracts, and flag unanswered messages. These are useful applications, but as Carlos Almansa Ballesteros, CEO of Nexudus, pointed out during Coworking Tech Week, the real value appears when AI understands coworking context — bookings, memberships, meeting rooms, support tickets, pricing, access control, events, and the rhythm of a physical space. Generic tools can help, but coworking operators need AI that speaks their language.

Much of what is marketed as "AI-powered" in the industry is still a rebrand of basic automation. The practical use cases that actually work today are more specific: drafting member communications, analyzing access frequency and booking patterns to predict churn, routing support tickets, and generating community content. These are real, measurable wins. But the bigger opportunity appears when AI can work with the accumulated memory of the business.

A coworking space produces valuable information every day. A member mentions that they may need a larger office. A recurring maintenance problem appears in support messages. An event generates several promising conversations. A pricing exception is approved. A company signals that its team may be growing.

This knowledge is usually spread across inboxes, calendars, chat messages, CRM records, accounting software, access data, meeting notes, and the memories of individual team members. As James Brouard, Co-Founder of Hamlet, described it: a community manager should not have to ask a colleague what happened with a member yesterday if the system can summarize the relevant context. A site lead should not have to rebuild the week from scattered messages if software can surface what needs attention.

With the right structure and permissions, an AI agent could review this information overnight. It could connect decisions with outcomes, identify recurring patterns, surface unresolved commitments, and prepare relevant context for the team. Platforms like Nexudus and Spacebring are already moving in this direction — building AI layers that work from the operator's own live data rather than disconnected exports.

Imagine the AI notices repeated complaints about a meeting room. It reviews booking data, maintenance records, and member feedback. By morning, it has prepared a summary of the issue, identified when it started, listed the affected members, and suggested the next actions for the team to review.

The daily output could be a short morning briefing:

• Members who may need attention • Leads that are losing momentum • Commitments that remain unresolved • Operational issues that keep returning • Opportunities to expand a service or account • Decisions that require human approval

Over time, this could support measurable improvements in member retention, occupancy, lead conversion, response times, event performance, and operating costs. The best argument for AI in coworking, as the Hamlet session put it, is not that it makes spaces more technical — it is that it takes enough repetitive work off the team's plate for people to be more present in the space.

The quality of the result depends on the quality of the business memory. Decisions need to be recorded. Outcomes need to be connected to them. Information needs clear ownership, permissions, retention policies, and audit trails. Sensitive member, payment, and access data require especially careful handling. Manuel Conti, CEO of PONT, made this point clearly: if the current data is inconsistent, manually reconciled, or split across too many tools, AI will produce more noise than insight. The first job is visibility. Intelligence comes after that.

The practical path starts small. Spacebring's Helga Moreno outlined a 90-day approach that many operators could follow: month one, focus on member support — let AI handle repeated FAQs, ticket routing, and simple booking questions. Month two, move to community and events — use AI to draft announcements, prepare event promotions, and summarize member feedback. Month three, bring in revenue insight and decision support — identifying underperforming products, forecasting availability, and highlighting opportunities the team might miss. First, the AI summarizes what happened. Then it identifies patterns. Once the underlying information becomes reliable, it can begin supporting forecasts and recommendations.

This creates a coworking business that learns from its own experience. Each morning, the team starts with clearer priorities, stronger context, and a better understanding of what the business has been trying to tell them. The future is not a teamless coworking space. It is a calmer, more proactive community team — with more time available for the human work that makes coworking valuable.

If your coworking space could dream tonight, what would you want it to understand by morning?