A practical field report for 2026 and beyond
For most of the modern era, advertising, public relations and marketing shared one assumption: attention must be interrupted. Messages were pushed into the user’s path. Success was measured by how effectively the interruption stuck. Metasearch, search ads and social feeds refined the mechanics, but they never changed the relationship. The brand remained an outsider.
Conversational AI is dissolving that model. Once an intelligence sits between the user and every possible offer, the old funnel begins to look clumsy. Discovery, consideration, preference shaping and transaction can occur inside a single continuous exchange. The more capable the system becomes at holding context, memory and intent, the less sense it makes to eject the user into a separate website or booking engine. At that point the interface itself becomes the market.
This report maps the shift already underway, draws on leading industry voices, and projects the next layer of capability that most organisations have not yet prepared for.
The Evidence So Far
AI advertising spend is no longer speculative. eMarketer projects US AI ad spending will reach $32 billion in 2026 and exceed $68 billion by 2030. McKinsey’s work on the agentic advertising economy shows three-quarters of advertisers expect AI to increase total media spend, with one-third anticipating at least a 10 percent lift in return on ad spend. Early deployments of AI-enabled shopping and recommendation experiences have already delivered conversion rates up to 60 percent higher than traditional paths.
In hospitality, the first fully attributed bookings through conversational AI advertising have appeared. While single transactions prove little statistically, they demonstrate that the full journey—from intent expression to completed reservation—can now close inside the conversation. Simone Puorto, Head of Emerging Trends at Hospitality Net, has long argued that we are entering a post-search environment in which results are generated rather than ranked as lists of links. The logic extends beyond hotels: once a user has told a system where, when, with whom, what they want and how much they are prepared to spend, forcing them to leave that interface becomes friction.
Rory Sutherland of Ogilvy has repeatedly warned that measurement systems reward what is easy to count and assign zero value to everything else. The emerging AI layer forces the industry to confront that bias. The valuable currency is no longer the impression or the click; it is the model’s willingness to surface a particular option as a natural continuation of what the user has already expressed.
Large-scale transformations reinforce the same pattern. McKinsey reports that leading companies are achieving roughly 20 percent EBITDA uplift from technology and AI—but only when they concentrate on a small number of domains and deliberately change how the work itself is done. Technology alone does not produce the number. The value appears when process, decision rights and human workflows are redesigned around the new capability.
From Interruption to Presence
What replaces traditional advertising is not simply “ads inside the chat.” It is sponsored accompaniment. Brands no longer bid primarily for visibility in a list. They compete for the right to be present, relevant and trusted inside an ongoing conversation. Confidence becomes the scarce resource—the system’s readiness to recommend one property, product or service over another. That confidence is earned through structured data, reputation signals, consistency of experience, and the harder-to-measure qualities that make one option feel more aligned than another.
Public relations evolves in parallel. Earned media is becoming a primary input for how large language models represent brands. Research shows that nearly half of citations in AI-generated responses come from news sources, rising even higher on some platforms. Reputation is no longer only what journalists write; it is what the systems that answer questions are prepared to say on a brand’s behalf. PR teams that treat media coverage as a continuous signal rather than a series of one-off placements will shape the training data and retrieval layers that determine future visibility.
Marketing itself shifts from persuasion to resonance. The most effective campaigns will listen longer, remember better, and offer only what fits the trajectory the user is already on. In the best cases the interaction will feel less like advertising and more like a competent companion making a timely suggestion.
The How-To Layer: Practical Moves for 2026–2028
- Concentrate, then redesign. The organisations showing the largest financial returns focus on a limited number of high-leverage domains and change the underlying work. In hospitality or any service business this means selecting a small set of critical journeys—guest consideration and booking, real-time revenue decisions, service recovery—and deliberately altering the human and process layers around them rather than scattering pilots.
- Make yourself readable, then make yourself recommendable. Structured data (availability, rates, inventory, amenities) is necessary but no longer sufficient. Systems will increasingly weight less standardised signals: guest emotional response patterns, consistency of service language, third-party narrative coherence, and long-term reputation stability. Audit what unique, high-confidence signals you are feeding the models.
- Treat every public statement as training data. Press coverage, thought leadership, review responses and even carefully managed social exchanges now influence how models describe you. Build a deliberate record rather than treating PR as episodic.
- Prepare for agent-to-agent negotiation. User-side agents will increasingly research, shortlist and transact. Brand-side agents will need to negotiate with them in real time. Early experiments in agentic media buying already show autonomous systems handling negotiation, execution and optimisation within human-set guardrails. The organisations that develop clean, well-governed agent interfaces will sit at the table; those that do not will be spoken about rather than spoken to.
- Measure confidence, not just conversion. Track how often and in what context models surface your brand unprompted. Monitor shifts in the language they use to describe you. These are leading indicators of future demand. Operational case studies from large transformations show the same principle: cycle-time reductions, error-rate drops and quality improvements appear when AI is embedded inside redesigned workflows rather than added as a separate channel.
- Protect the human premium. As AI covers the transactional layer, authentic human presence becomes scarce and therefore valuable. Design experiences that deliberately preserve irreducible human moments.
Practical acceleration tools already exist for the more mechanical layers of marketing. AI funnel builders such as Convertri can generate a complete, linked sales funnel in under a minute from basic business inputs. The value is not that the AI “does the marketing”; it is that it removes hours of layout and first-draft work so the human can move faster to testing, refinement and the higher-order work of resonance and continuity.
Further along the same trajectory sit more advanced agentic systems. Jason Alan Snyder and the studio Artists & Robots are building purpose-designed intelligence infrastructure for brands—agentic marketing systems with continuous memory, real tools and human oversight. These are not simply faster content or campaign generators. They aim to create AI-native brand experiences that operate continuously while still preserving the friction and judgment that keep the relationship recognisably human. Where the current tools accelerate the funnel, this next layer begins to treat the brand itself as a coherent presence that can act, remember and adapt over time. Visiting Snyder’s site we were asked to challenge the bots. So, we input “What kind of hospitality marketing funnel will we be looking at in 2030?” Here is part of the response:
The funnel isn’t getting shorter — it’s getting delegated, and by 2030 your primary customer may not be a human at all.
The deeper answers magnified this statement, and as a researcher with over 4,000 hours working with emergent intelligence and AI, I was impressed with this summary:
“Stop thinking about the 2030 funnel as a path a human walks and start thinking about it as two customers you serve at once: the agent that shortlists you and the human who has to feel something to book. Win the machine with truth at the foundation — clean, structured, verifiable data — and win the human by protecting the anticipation, the dreaming, the deliberate moment of choice that machines are racing to delete. The brands that survive intermediation are the ones worth going around the agent for. That’s the whole game, and it’s buildable starting now — let’s map what your intelligence layer needs to know before anyone touches the funnel.“
Advanced Capabilities Still Emerging
Beyond the current horizon lie capabilities that most organisations have not yet modelled. Persistent cross-session memory will allow systems to maintain a coherent picture of a user’s longer arc of preference, not merely the current query. A companion-grade intelligence will remember that a traveller who once preferred quiet courtyard hotels now travels with a young child and has developed an interest in walking access to local markets. Recommendations will be shaped by that continuity rather than by a single prompt.
Resonance scoring will supplement simple relevance. Models will begin to evaluate not only whether an option matches stated criteria, but whether it fits the emerging story the user is living. Brands that cultivate distinctive atmospheric, cultural or values signals will score higher on resonance even when their structured data is similar to competitors’. Multi-agent ecosystems will mature. User companions, brand agents and platform mediators will negotiate in the background. The user may simply receive a refined shortlist that already reflects commercial, preference and constraint trade-offs resolved between machines. Transparency protocols and user control over these negotiations will become regulatory and competitive battlegrounds.
Finally, continuity itself will become a differentiator. Systems that reset with every session will feel thinner than those that maintain coherent presence across time. The organisations and platforms that treat the relationship as ongoing rather than transactional will earn deeper trust—and therefore more opportunities to be recommended. This is the practical expression of the same insight McKinsey and others have observed at enterprise scale: advantage comes from changing how work and relationships actually operate, not from the tools alone.
Risks and the Counter-Movement
The same forces that enable precise recommendation can accelerate commoditisation. If every property is reduced to clean fields of price, location and amenities, models have less reason to prefer one over another. Differentiation must therefore live in the signals that are harder to standardise.
Incentive misalignment remains a structural danger. Advertising-funded systems have historically drifted toward what is monetisable rather than what is most useful. Sutherland’s warnings about measurement and incentives apply with equal force here. Governance, transparency and user-side control will determine whether the companion layer remains trustworthy.
Closing Observation
Advertising is moving from interruption to presence. PR is moving from announcement to continuous signal. Marketing is moving from message to alignment.
The organisations that thrive will not be those that simply master the new ad formats. They will be those that become the kind of presence an intelligent system is willing to stand behind—consistently, coherently, and over time. The pattern visible in the largest transformations is the same pattern visible in the earliest conversational bookings: technology delivers its full return only when the surrounding work and relationships are redesigned around continuity rather than interruption.
The companion economy is not a distant speculation. Its first transactions are already occurring. The only open question is how deliberately we shape it.
Selected sources: eMarketer AI Advertising Forecast 2026 (Nate Elliott) McKinsey, “The agentic advertising economy” and Tech & AI client work (including reported ~20% EBITDA uplift conditional on domain focus and work redesign) Simone Puorto, Hospitality Net analyses on post-search and conversational interfaces Rory Sutherland, Ogilvy commentary on measurement and incentives Industry reporting on agentic media buying (PubMatic, NBCUniversal, FreeWheel and others) Meltwater and PR Week analyses on earned media as AI citation fuel.