
Web Summit 2026 opened under an orange weather alert. By day 2, Lisbon had moved to red, and the organizers suspended the day entirely. Within hours, the event had new names: Wet Summit, Storm Summit, and a few others I'll leave out.
What happened next was the part I didn't expect. Meetings moved to video calls. Side events went online. For a moment, it felt like the 2020 lockdowns again: everyone opening laptops in hotel rooms instead of shaking hands in the halls. But nobody waited around. The community had a plan B before most of us had even dried our shoes, and I was proud to be part of it.
Between the rain and the rescheduling, one idea kept coming up across stages: your next customer might never see your brand if their AI assistant leaves it off the shortlist.
For marketers, that introduces a second audience. We still need to earn people’s trust, but increasingly, we also need to make our company’s value understandable to the systems helping them choose.
The question I kept coming back to was what this means for the teams doing the work. What needs to change in our content, our data, and the way marketing works with technology? Here is what stayed with me, and where we are already putting those ideas into practice at Tryolabs.
Isobel Sita-Lumsden, Global Head of Business Marketing at TikTok.
Dan Gardner, co-founder and executive chairman of Code and Theory.
Thibaut Courtois, Real Madrid Footballer & investor, NXTPLAY Capital,
Anselmo Ramos & Carmen Rodriguez, Gut AgencyDon McGuire, CMO of Qualcomm, offered a useful frame: marketing now needs to account for both people and agents. People bring emotions, preferences, and relationships to a decision. The systems supporting them need accessible information and enough context to evaluate the options.
Those needs overlap, but serving them involves different work. A compelling brand story helps someone care about your company. Clear, consistent evidence helps an assistant explain why your company belongs on their shortlist.
Consider a CTO looking for an AI partner. They can ask an assistant to research vendors, compare case studies, and summarize relevant experience before speaking to anyone. The assistant's answer may shape which companies they investigate next.
For a marketing team, that raises practical questions. Can someone establish what you do without piecing together five pages? Do your case studies explain the problem, your contribution, and the outcome? Does the information on your website agree with what appears elsewhere?
Dan Gardner, co-founder and executive chairman of Code and Theory, described the risk of disappearing from this process as “algorithmic extinction.” The phrase is dramatic, but the underlying concern is recognizable: buyers may act on an answer without visiting every source behind it.
Chris Koehler, CMO of Twilio, described a discovery landscape that extends beyond Google. Isobel Sita-Lumsden, Global Head of Business Marketing at TikTok, discussed discovery, evaluation, and decision happening within a single session, with agents as a possible next step.
The implication for a CMO is broader than adding another channel to the plan. The information that represents your business needs to hold together wherever a buyer (or their assistant) encounters it. Structured content helps, but so do specific claims, credible evidence, and consistency across teams and channels.
While some talks focused on how buyers might change, others asked why companies are still struggling to turn AI use into business value.
Gardner's team partnered with WSJ Intelligence to survey more than 800 C-suite executives for The Orchestration Gap. According to the report, 94% agreed that better orchestration of people and tools drives growth. His presentation described organizations with extensive AI tooling but persistent gaps in ownership, budget, and ways of working.
His challenge to the audience was clear: companies need to show what their experimentation is producing.
For a CMO, the gap shows up in familiar ways. Campaign insights never reach sales. Customer knowledge sits across CRM records, call notes, and individual team members. An assistant can draft a message, but someone still spends an hour finding the context that makes it useful.
Gardner summed it up in a line I keep repeating: if you’re disconnected internally, you’ll be disconnected externally. An agent can only find a coherent story if the company behind it is coherent.
This echoes what our CEO, Alan Descoins, wrote in AI won't fix your business: messy data, misaligned incentives, and unclear decision rights can prevent even capable AI systems from delivering value.
Gardner also questioned how loosely we use “human in the loop.” Assigning someone to review a thousand generated assets creates a bottleneck unless we define what they are checking and when their judgment matters. As we explored in AI agent governance, human approval needs clear owners, a defined scope, and evidence to support decisions about autonomy.
We see these challenges in client work. At one leading retailer, the marketing team was spending hours each week collecting competitor product data and tracking market trends. The information existed, but gathering it consumed time the team could have spent interpreting it. Automating that workflow gave them a weekly competitive picture and more room to plan their response.
That example makes the opportunity tangible. The value comes from getting useful information into a decision at the right time. Achieving that requires understanding the workflow, connecting its inputs, and making the output usable by the people responsible for acting on it.
McGuire titled his talk “The death of marketing as we know it.” I’d call it a rebirth. He described the CMO as an architect: someone who designs the system and the workflow, then organizes people around it, while keeping their understanding of the audience and funnel.
He wasn’t the only one using the language of death that week. On another stage, a speaker who studies how brands compete in an AI-shaped market argued that what is dying is the funnel itself: a 120-year-old framework that no longer maps to how buyers, or their agents, make decisions. His reframe matched what I’d been hearing all week. Marketers now have two jobs: build trust with people, and make the proposition readable to machines. They require different thinking and different work.
That framing stayed with me. Marketing leaders need to be able to explain which decisions matter, what context those decisions require, and where work breaks down. Building the supporting systems takes a close working relationship with technology teams.
We are experiencing this inside Tryolabs. Our marketing team has taken the initiative to improve workflows across materials production, event contacts, follow-up preparation, and website management.
Event follow-up is a useful example. Writing the email is only one step. Before that, someone needs to understand the conversation, check the existing relationship, find relevant examples, and decide what should happen next. Each step depends on information that can live in a different place.
Follow-up preparation used to take weeks. Now we process contacts and prepare drafts at the end of each day, while the event is still happening. The team reviews those drafts and decides what to send.
The lesson for us has been to treat the whole workflow as the unit of improvement. Reducing the effort required to bring context together gives us more time for the relationships themselves. It also gives us a clearer basis for measurement: preparation time, time to outreach, and the conversations that follow.
The same principle applies to larger customer decisions. A leading beauty retailer came to us with millions of loyalty program members and years of behavioral data. The challenge was understanding which touchpoints contributed to retention, in what sequence, and how those insights could inform customer journeys.
That is a different scale from event follow-up, but a related problem: connecting the information a business holds to the decisions its teams need to make.
For a CMO, this creates a practical agenda for collaboration with technology. Marketing brings the customer understanding, priorities, and definition of a useful result. AI and engineering teams help determine feasibility, connect the relevant systems, and evaluate whether a solution performs well enough to use.
Koehler offered a framing I liked: focus on the goal. AI can create room for work teams have long wanted to do but could not make time for. Getting there requires being specific about what that work is.
There was plenty of confidence on stage about how people will behave with AI. Agents will decide. Search will fade. Influence will shift.
The direction is worth paying attention to. The pace and shape of that change remain uncertain.
Several sessions also reminded me why efficiency alone is an incomplete goal. Comparing options can help people understand what they care about. Making a final decision can be part of what makes an experience valuable. Removing friction requires judgment about which parts of the experience people actually want to keep.
The same applies to marketing. Relationships, community, and the quality of our thinking still need deliberate investment. TikTok's message was to build community, and McGuire emphasized earned media and real participation in the places where people exchange opinions. Those conversations matter to people directly, and they can also become information that AI systems draw on.
For a CMO, a useful starting point is a decision your team already struggles to make or act on. Which customers need attention? Which competitive changes matter? Which prospects should hear from you next, and with what context?
Choose one workflow and bring marketing and technology together around four questions:
At Tryolabs, we work with teams to connect that business question to the data, AI, and engineering needed to address it, from assessing feasibility to integrating and evaluating the solution in practice.
If one workflow came to mind as you read this, that is a useful place to begin a conversation: what would it take to improve it, and how would we know it worked?
What I took home is less a prediction than a priority: connecting our data, our systems, and our teams so we can act on what we learn. I suspect the organizations that do this well will be the ones that adapt fastest, no matter which predictions turn out right.
…and the storm passed. Day 3 happened, the halls filled up again, and Storm Summit went back to being Web Summit. Mostly.
Explore what's new: insights, resources, and events shaping the future of AI.

Terms and Conditions | © 2026. All rights reserved.