We just got back from Quirks London 2026. Two days, dozens of sessions, and one general conclusion: the insights industry is entering a period of real change. Not the kind where everyone agrees on where things are heading—the kind where the debate itself tells you something important.
Here are the five trends that kept coming up, on stage and off.
1. Research Agents Are Moving from Assistant to Orchestrator
A year ago, “AI in research” meant a tool that helped you write a survey or summarise open ends. At Quirks London 2026, the conversation had shifted meaningfully. The term gaining traction is research agent — an AI system that doesn’t just assist with a single task but manages entire multi-step research workflows end-to-end.
E-Tabs presented alongside MMR on exactly this: combining AI-powered product testing with automated reporting to create what they called a fully integrated, end-to-end insights engine. Across the event, figures like a 50% reduction in production time were cited as what full pipeline automation can deliver. The question is no longer “can AI help?” but “what happens to research quality and team capacity when AI handles the whole pipeline?”
For us at CodexMR, this is the conversation we have been building toward.
2. Brands Are Actually Running the AI vs. Human Experiment
One of the most talked-about sessions came from Anthropic, who presented findings from running focus groups and AI interviews side by side to research their own marketing. That is a remarkable thing to do, and an even more remarkable thing to share publicly at an industry event.
Arla followed with a session on making AI-led interviews the primary vehicle for consumer listening – framing it as “removing the excuses for not listening to consumers.” Both sessions landed because they were not theoretical. These are brands that ran the experiment, got data, and came back to say what they found.
Our takeaway? Neither side declared a winner. What emerged was more nuanced: AI interviews excel at speed and scale; human methods carry weight for emotional depth and complex behavioural questions. The smartest teams are learning to choose deliberately rather than default.
3. Insights Activation Is the New Battleground
L’Oréal’s session reframed something the industry often glosses over. The title said it plainly: Make insights impossible to ignore. Their focus was not on how to collect better data — it was on how to make stakeholders actually act on what research teams already know.
Sky and BayesPrice pushed the same idea from a different angle, talking about moving ownership from “the data point” to “the point of view.” Research teams are being asked to do less reporting and more advising. The data is almost never the problem. The story told with it usually is.
This is showing up in the broader industry data too. Across multiple 2026 research reports, the consistent finding is that teams actively using agentic AI report significantly higher efficiency – and that the freed-up capacity is going toward strategic advisory work, not more fieldwork.
4. Synthetic Respondents: Big Money, Bigger Questions
The investment flowing into synthetic research is impossible to ignore. Simile, a Stanford spinout building AI digital twins from real interview data, raised $100M in February. Listen Labs raised $69M in January. Aaru, which generates entirely synthetic AI populations, hit a $1B valuation. This is not fringe activity.
And yet, Rival Group’s 2026 trends report found that 42.75% of researchers are still “not excited” about synthetic respondents. The scepticism is not irrational – validation data is mixed, methodology is still maturing, and the use cases where synthetic respondents genuinely perform well (pricing, ranking, structured concept testing) are quite specific.
What struck us at Quirks was that the sharpest practitioners are not taking a binary position. They are asking: for this specific research question, at this moment in the project, what is the right tool? That is the right frame.
5. Speed Is No Longer a Nice-to-Have
Perhaps the most consistent undercurrent across all two days was urgency. Brands are not waiting weeks for research anymore. The phrase we kept hearing — “insight at the speed of business” – captures something real. Decisions are being made faster, stakeholders expect input in hours not weeks, and research teams that cannot keep up are being cut out of the conversation entirely.
This is not just a vendor pitch. It is showing up in how brands are restructuring their insight functions and what they are demanding from their suppliers.
Wrap Up
Quirks London 2026 did not give us easy answers. What it gave us was a clearer picture of where the pressure is – on speed, on activation, on the credibility of new methods. Those are exactly the problems worth solving.
Our read, having spent two days in those rooms, is that the companies pulling ahead are not the ones who have adopted the most AI tools – they are the ones who have figured out where AI execution ends and human judgment begins.
The debate between automation and expertise is a false one. The real edge is in combining AI’s ability to move fast and at scale with the kind of expert oversight that keeps the research decision-ready and commercially grounded. That is exactly the model we have been building at CodexMR: QuantOps-powered delivery with researchers who know when to trust the output and when to interrogate it.
If Quirks sparked some thinking about where your own research stack is heading, we would love to continue the conversation.
Our commercial lead Stefka Mihaylova is available at [email protected] and you can book a call directly at our website to talk through what we are doing and how we can help you move the needle.