Based on existing data and early experiments (e.g., ChatGPT ad tests and startups like Nexad/Dappier), what revenue ranges are realistic when embedding ads directly in AI UI flows, say, within conversational agents or assistant interfaces? Can ads inside AI solutions alone cover ongoing compute costs and profitability targets?
At this stage, AI-embedded ads look promising, but they’re not a silver bullet. What we’re seeing across performance marketing is that high-intent, context-aligned placements consistently command better engagement. Whether that’s search ads, retargeting, in-app placements, interstitials or the new wave of conversational environments.
But even with premium CPMs, it’s unrealistic to expect advertising alone to pay for the enormous compute bills behind frontier‑scale AI. We’ve seen the same dynamic in other fast‑growing channels, like Telegram Mini Apps, where performance is strong, but infrastructure growth still requires multi‑stream monetisation.
Academic research shows ads inserted in conversations manipulative once disclosed. From a product design and retention perspective, what are the real limits of ad saturation inside AI experiences without degrading trust or usage?
For us, trust has become the defining trend of the last two years. In 2025, we saw advertisers and users staying away from anything that even smells manipulative. That’s why clean traffic, transparent logs and strict anti-fraud protections became non-negotiable. Once an ad starts feeling like advice, you’ve already lost the user.
So the red line is simple: ads must stay clearly separate from the assistant’s voice. If the line blurs, trust collapses and no amount of contextual relevance can save the format. In our world, sustainable formats are the ones that don’t pretend to be something they’re not, and where the match between the audience and the offer remains the key driver of performance.
What models of in-product advertising are being tested today (e.g., bottom-of-response ads, contextual placements, sponsored prompts), and which models have shown measurable traction without compromising the core AI experience?
Across all the tests we’ve run in emerging channels, the formats that win are the ones that replicate what already works – high intent and clearly separated sponsored units, such as in app ads, interstitials and now Telegram Mini App placements. These all deliver because they meet the user at the right moment without hijacking the experience.
Conversational ads should follow that same logic. Adjacent placement works; in‑conversation steering does not. Anything that interrupts the natural flow of a chat risks immediate drop‑off.
If advertising becomes a substantial revenue line in AI solutions, what are the foreseeable impacts on adjacent markets (e.g., search ads, enterprise AI pricing, consumer expectations, regulatory scrutiny)? What externalities should business and policy writers be aware of?
If conversational AI becomes a large-scale ad channel, it’s going to reshape the funnel the same way push notifications and pop formats once did. Discovery will shift further upstream and brands will optimise for conversation state, not keywords.
For performance-oriented platforms, this means immediate adaptation. We will need to rethink ad formats, client strategies and, more fundamentally, where we sit in the buying funnel as conversational AI reshapes the user journey. The experience inside AI tools is different, and advertising strategies will have to evolve with it rather than simply replicate legacy models.
At the same time, traffic quality, fraud prevention, and results-based optimization will always remain relevant regardless of the channel.
Looking at wider trends, what will ads in ChatGPT mean for retail media?
There’s been a lot of speculation about what ads in ChatGPT could mean for retail media networks, but it’s unlikely to be a zero-sum shift. Retail media works because it sits close to the point of purchase, where intent is already strong and conversion happens. ChatGPT advertising would operate much earlier, during the consideration phase when users are asking questions like ‘what’s the best…’ or ‘help me choose between…’, shaping decisions before a user even lands on a marketplace or retailer site.
If anything, this could increase the value of retail media by sending more qualified traffic into those environments. The real impact will be felt in how brands plan budgets and messaging across the journey. Advertisers will need to think more carefully about which channels influence consideration and which close the sale.
The bigger change is behavioural. Discovery is becoming conversational, rather than navigational. Users aren’t just searching and clicking; they’re asking, refining, and deciding within the conversation itself. That doesn’t weaken retail media networks, but it does mean they’ll exist within a more complex, multi-touch advertising ecosystem.
For advertisers, the question isn’t ‘ChatGPT or retail media’, it’s ‘how do we build a strategy that recognises where each adds value, and ensures our brand shows up at both moments that matter for final conversion’.
Final thoughts on AI in advertising
In 2026, AI will stop being treated as a performance “boost” and become the operating system behind modern advertising. Ecommerce is moving toward marketplaces and app-based shopping, which means advertisers are juggling way more variables: product, price, device, GEO, creative, timing. No team can manage all of that manually in real time anymore. AI is stepping in to handle these micro-decisions – reallocating spend, testing creatives, adjusting campaigns as conditions shift.
Human teams aren’t going away, but their role is changing. The best advertisers will use AI to take care of day-to-day optimization so teams can focus on strategy, product selection, and long-term growth. In a more competitive market, AI will separate the advertisers who scale from the ones who stall.
