The Rooms Where People Could Actually Meet
A storm, too many AI pitches, and the conversations I nearly missed in Lisbon
November 15, 2026Read on Substack
I arrived in Lisbon last Saturday.
By Wednesday, the days had taken on a familiar conference rhythm. I would walk to the next venue, find the gathering, take a moment to work out who was there, and start talking to people. Sometimes an introduction stayed an introduction. Sometimes a conversation became interesting just as the next event on my calendar approached. I had deliberately left myself room to choose whether to move on or stay.
Then Web Summit cancelled its Wednesday program after the Portuguese government advised against holding it because of risks from Storm Charles. Everyone who had built the day around the main venue suddenly needed a different plan.
Some satellite gatherings were postponed or moved online. Others went ahead. By the evening, I was at an event where people whose conference plans had fallen through kept arriving. The gathering was becoming different from the one its organizers had originally planned: the room was changing as word of mouth reached more people. Seeing this, the organizers extended the program to accommodate them instead of leaving them out.
The main venue was closed, but some of its displaced crowd had found its way into the spaces that remained open.
The change also gave me time to sit down with . During my last years in New York we used to run into each other regularly, but I hadn’t seen him since moving to Europe. We had both reinvented ourselves in the meantime. Instead of catching up in passing, we could talk properly about life and the years in between. I gave him a copy of The Human OS Manual, and we left with a few ideas worth returning to later.
That one conversation was valuable to me, but the cancellation was a safety decision, not a stroke of good fortune. Organizers had to act on the risks as they understood them, and exhibitors lost a day they had paid and prepared for. What stayed with me was how quickly the circumstances of meeting someone could change.
The people hadn’t disappeared. They were gathering elsewhere, in different numbers, with different expectations of what the day might bring.
An Invitation Is Not An Encounter
Web Summit reported 70,003 attendees and 1,943 investors for its 2026 edition. That kind of concentration is why a conference like this exists. You can encounter founders, journalists, researchers, and potential partners in a few days who might otherwise be spread across several countries and months of travel.
Holding an investor pass opened a different part of the conference: smaller gatherings, dinners, private events, and introductions outside the main venue. Some of my more substantive conversations happened there. But the badge also brought a steady stream of connection requests, many with little apparent relevance to my interests or the work I had come to do. A few were thoughtful and specific. Others looked like pitches sent to every investor whose name appeared in the conference app. Each request was another small decision about where to spend my attention.
Years of attending conferences, and of organizing side events myself, have taught me to treat the schedule as a set of informed choices rather than a fixed itinerary. I select gatherings that are genuinely relevant, even when some overlap, then decide on the ground where my attention is best spent. I try to release an RSVP when I know I won’t attend, especially when capacity is limited. Having organized events, I know that not every registration turns into an arrival. I described this approach in a recent Medium article before coming to Lisbon.
If a conversation is going somewhere useful, I can stay. If a session offers little reason to remain, I can leave. That is a practical advantage of keeping the schedule flexible, although it still depends on finding the right people in a room where there is time to talk.
When Web Summit resumed on Thursday, the postponed sessions were being folded into an already crowded schedule. I decided not to return to the main venue and visited another technology gathering instead. The sessions were small; outside them, the common areas were almost empty. Often, the useful part of a conference begins after a talk, when someone stays to finish a conversation or introduces you to someone else. Here, there were hardly any such opportunities. There may have been interesting people in the sessions, but very few stayed around for the spontaneous encounters I had gone there hoping to find.
At one venue, introductions were competing for attention. At the other, there were barely enough people around for an unplanned conversation to begin. Neither experience told me much about the quality of the companies or people present. But I could feel the difference between a room where it was hard to stay with someone and one where there was almost nobody to meet.
The smaller gatherings and dinners worked differently. They provided enough shared context to begin talking, and often enough time to get past the name, title, and prepared explanation of what someone was building. The room did not have to be quiet or intimate. People simply had to be able to stay with a conversation.
What The Next Conversation Inherits
During a rapid sequence of startup pitches, I was trying to attach a particular business to each name. A founder had just a minute to explain what was being built, whom it served and why it deserved attention. Then the next founder stepped up. “AI” in the company name, “AI” in the website address, “AI” in the opening sentence. A few minutes later, there it was again.
At first, the repetition was merely noticeable. After enough pitches, hearing “AI” told me almost nothing about which company was speaking. There is a term, semantic satiation, for the way a word can temporarily lose its meaning through repetition. I wasn’t measuring anything, but the analogy was close to what I experienced. And beneath the repeated label, I kept hearing propositions that sounded less like independent businesses than features already available through general-purpose assistants. The branding made the companies harder to distinguish; the overlap in what they offered made the problem more consequential.
One pitch stayed with me because it offered a product for organizing email, calendars, and administrative tasks. I already do much of that with my out-of-the-box AI assistant. As I listened, I found myself adding up what a customer would be taking on: another interface, another provider, another cost. There might be a good reason for all three, but the presentation had not yet given me one.
There could be good answers. A dedicated product might work within an organization’s security requirements, connect to difficult legacy systems, or solve a particular industry’s problem far better than a general-purpose assistant. I have spent enough years building and selling enterprise software to know that a feature can become a business when the harder work around it has been done.
That difference wasn’t apparent in the pitch. A few minutes onstage is not due diligence, and I may have missed something important. Still, a useful feature is not automatically a defensible business. The more versions of a familiar capability I heard, the harder I had to work to find the genuinely different proposition.
Somewhere during the sequence, I noticed a change in how I was listening. I was no longer asking first what a founder had seen that I hadn’t. I was anticipating the weak point before the explanation was complete. My bullshit detector had taken over.
An investor needs that reflex. Time, attention, and capital are limited, and founders should expect scrutiny. The trouble is that suspicion can get ahead of judgment. The next founder might have a genuinely difficult problem, a credible advantage and the ability to execute, but they still have to get through the impatience left by the pitches before them.
Later I came across an analysis by Feedance, which was also exhibiting at Web Summit. It had examined public descriptions of 2,587 startups and classified 696, or 27 percent, as offering core functions that AI skills and standard connectors could largely cover. The exercise used automated keyword rules, not product testing. It cannot establish which companies have viable customers, defensible integrations, or businesses worth investing in. But it did offer some corroboration for the overlap I thought I was hearing earlier.
The cost of that scrutiny does not end with the pitch session. The next person you meet inherits whatever patience, suspicion, or curiosity you have left. They may have something original to show, but they’re talking to someone who has already spent an hour separating plausible businesses from familiar promises. Neither person can see all of what the earlier conversations have done to this one.
Conferences count introductions and meetings. They have little way of accounting for what the previous ten conversations did to the eleventh.
What Changed When I Left The Conference Circuit
By Thursday, I’d spent much of the week in places where someone was explaining a company, making an introduction or deciding whether a conversation should continue. That afternoon I joined a Founder Decompression gathering at a venue built around heat and cold. I teach breathwork and contrast exposure at my retreats, but here I was a participant. For a couple of hours, I wasn’t responsible for guiding anyone else through the experience.
The sequence itself was familiar: sit with the sauna heat, go into the cold, then find yourself talking with people between practices. Heat and cold ask for a kind of attention that another presentation does not. For once, I wasn’t guiding the experience or wondering whether a more useful conversation was happening in another room. I could take part in what was happening in front of me.
I ended up talking at length with a team working in coaching and community building. Our work overlapped, but the conversation didn’t begin with a presentation or a request. We talked about what each of us was doing, where the interests met, and what collaboration might look like. There was no reason to turn it into an agreement before we left. We could simply keep talking.
Later I attended a longevity dinner. On the calendar it was another gathering, with more people to meet and conversations to begin. But I felt noticeably more ready to be there than I had earlier in the day.
At dinner I noticed that I wasn’t trying to decide quite so quickly whether a conversation was worth pursuing. Earlier in the week, screening claims and protecting my time had become almost automatic. I was still interested in what people were working on, but I could give the exchange more room before deciding what it meant.
I cannot isolate how much of that change came from the heat and cold. It may have been the pause, the format, the conversations, or simply a couple of hours in which I was not performing a commercial role. I did not measure any of it, and I have no reason to assign the effect to a single cause.
The Part The Attendance Figures Miss
Attendance numbers have a legitimate purpose. An event has to bring people together before any meeting can happen, and organizers need to understand its reach. Web Summit’s scale is part of what makes a trip like this worthwhile.
But attendance tells us very little about what happened after two people were introduced: whether they had time to understand one another, whether an investor was still receptive, or whether a conversation survived the walk from the exhibition floor to somewhere quieter.
I would not turn this into an argument for smaller conferences. A large gathering makes encounters possible that a dinner of twelve never could. Nor is it a reason to abandon planning and trust chance. I came with a plan, and that is partly why I was able to respond when the plan changed.
A calendar can tell me where I’m supposed to be. It cannot tell me what kind of attention I will have left when I get there.
Consider the founder who finally gets eight minutes with an investor after the investor has spent an hour hearing variations on the same promise. On paper, the introduction happened. In practice, the founder is also dealing with the residue of conversations they were never part of. The same can happen when a journalist, collaborator or prospective customer arrives with a day of other people’s demands already behind them.
I went to Lisbon prepared to choose among a great many rooms. I would prepare much the same way again. But some of the conversations I valued most happened when I had enough time and attention to stay with the person in front of me.
The calendar helped me get there. It could not tell me in advance which conversations would matter.
Notes and sources
- Web Summit (2026). “Web Summit 2026: Record number of investors at its most international event yet.” November 10. Organizer-reported attendance and investor figures.
- Davies, P. (2026). “Web Summit Lisbon day 2 cancelled due to Storm Charles.” Euronews, November 10. Government safety advice and announced cancellation.
- Yılmaz, K. (2026). “Web Summit 2026 in Review: A Free Guide to All 2,587 Startups, and What They Tell Us About AI.” Feedance, November 12. Its keyword-rule classification of public descriptions is not independent product testing or a measure of business viability.
- “Causes, consequences, and strategies to deal with information overload: A scoping review.” International Journal of Information Management Data Insights (2024). General context only; not evidence of changes in attention at Web Summit.
- American Psychological Association (2018). “Semantic satiation.” APA Dictionary of Psychology. Defines the apparent loss of a word’s meaning after rapid repetition.