A giant screen, a conversational interface, a promise of productivity. A few booths away, almost the same scene plays out again. For companies attending VivaTech, CES or the RAISE Summit, the challenge is no longer simply to showcase an innovation: it is to give people a reason to stop, then to believe in it. Amid the flood of announcements, evidence becomes a communications tool. Provided it can be examined, explained and discussed without everything depending on the demonstrator’s skill.
Looking ahead to September 2026, this battle for attention deserves to be approached without confusing established facts with predictions. The following analysis draws on trends documented since the rise of generative AI, particularly at major events in 2024 and 2025. These forward-looking recommendations are not a review of the 2026 editions: they outline a strategy for companies facing persistent media noise.
Three events, three ways to lose your audience
These events are not interchangeable. In Las Vegas, CES remains defined by products, devices and demonstrations: mobility, smart homes, health and computing. An innovation often needs to be understood within seconds. In Paris, VivaTech brings together major corporations, startups, investors and public-sector decision-makers. Announcements sit alongside the search for partners, recruitment and corporate communications.
The RAISE Summit, launched in Paris in 2024, is more directly rooted in the artificial intelligence ecosystem and enterprise adoption. Discussions there can be more specialized, but they still face competing narratives. In all three cases, arriving with the same generic press release means ignoring what each audience is looking for: a tangible product, a credible partnership or an operational solution.
Since ChatGPT’s release in late 2022, AI terminology has become ubiquitous. Assistants, copilots, agents: the words change faster than the available evidence. CES 2024 notably illustrated how this label was spreading to computers and consumer devices. The implication for communications professionals: the presence of AI is no longer enough to make something newsworthy. Companies need to explain what it changes, for whom and under what constraints.
A demonstration that welcomes scrutiny
A good demonstration does more than show that something works. It helps people understand why the result deserves attention. Take an assistant tasked with handling after-sales service requests. Displaying a fluent response proves very little. Showing the documents consulted, the relevance of the response, the time required and the situations in which the system hands over to a human already begins to tell the product’s story.
The protocol matters as much as the interface. Was the scenario prepared in advance? Are the data real, anonymized or synthetic? Does the response come from the system being presented or from a recorded result? A backup video is perfectly legitimate at a trade show where the network may fail. But it must be identified as such. Transparency protects a reputation better than the illusion of absolute reliability.
Show the difficult cases too
Planning for a difficult case can become an editorial advantage. An incomplete document, an ambiguous question, an out-of-scope request: these obstacles reveal the safeguards. For a journalist or a buyer, a tool that knows when to refuse a task can be more interesting than one that always answers. The aim is not to stage a failure, but to make the conditions of use visible.
- Define the task: what concrete problem should the product solve?
- Explain the comparison: against which existing process or tool?
- Document the result: using what method and with what limitations?
- Enable verification: what will a third party be able to test or consult after the show?
The customer as a witness, not a prop
At a booth, a familiar logo catches the eye. It says nothing about the reality of deployment. The gaps between an experiment, a signed contract and everyday use are considerable. The role of communications is to clarify those gaps, not erase them. A credible customer advocate must be able to describe the starting point, the difficulties encountered and what has actually changed.
Their account becomes more compelling when it specifies the scope: a team, a department, a category of requests. An improvement observed on one task does not demonstrate an equivalent gain across the entire company. Quantified results, where available, must come with their observation period, their baseline for comparison and the human work still required. Otherwise, the figure becomes a fragile slogan.
There must also be room for candor. A conversation in which the customer can discuss integration, training or internal resistance provides more substance than a string of compliments. If no user can speak publicly, it is better to explain why and offer limited but robust documentation. Creating a false impression of maturity risks an embarrassing fact-check.
Prepare the spokesperson for uncomfortable questions
The spokesperson is not simply the person who recites the vision. They must connect a commercial promise to technical, economic and legal realities. Where do the data go? Who oversees the responses? What does intensive use cost? What happens if the model provider changes its pricing or terms? These are legitimate questions, even during a ten-minute interview.
Useful preparation relies on challenging rehearsals, not memorizing talking points. The executive must be able to distinguish between what works today, what is still being tested and what belongs on the roadmap. On European AI regulation, which is being phased in, it is better to specify the company’s role and the relevant obligations than to proclaim blanket compliance without evidence.
Plan a sequence, not a fireworks display
Standing out begins before the doors open. A brief press kit, sent to journalists with a genuine interest in the subject, can present a use case, access to the demonstration and an available contact. An embargo can help journalists prepare, but it must be negotiated: writing it in an email does not create an agreement. During the event, the quality of access matters as much as the quality of the messaging.
After the show, the company should be able to provide the promised materials quickly: measurement methodology, technical details, a customer interview. Evaluation also benefits from looking beyond the volume of mentions. An article that explains the product accurately, a qualified sales meeting or a request to test the product can count for more than a superficial rehash of the press release. It all depends on the objective set beforehand.
What now? At the next major technology events, announcements about agents and automation are likely to continue multiplying. What makes the difference could then come down to a very simple discipline: show fewer scenarios, but document them better; make narrower promises, but give more precise answers. For communications teams, the task is not to make complexity disappear. It is to make innovation clear enough to attract attention, and verifiable enough to keep it.
