Market research executives gathered in London yesterday for the AI: Powering-Up Insights conference 2026, where they discussed a range of issues, including what AI does for B2B insight, AI-moderated interviews, synthetic respondents, using AI for enhanced content metadata and why AI tools need better researchers.

Among the many case studies on show, some clear themes emerged, such as that AI was saving organisations time, but that having a human in the loop was still crucial with AI-centred research, which was still at the experimental stage for many organisations.

Representatives from oil giant Shell shared lessons from using AI across its global B2B insight projects, including how they were using the technology to synthesise complex datasets, refine segmentation and speed up innovation.

Pooja Amin,  industry + off highway insight lead at Shell, said AI “can pull together lots of information from our insight library and help us identify where the gaps are in our research and what sort of questions we might want to think about putting into our research”.

However, she urged caution about using synthetic respondents for collecting data, particularly when it comes to niche B2B audiences in areas like mining and agricultural audiences, as there is limited data to train AI models on.

She said that AI was saving Shell researchers at least 10 minutes a day. Amin said: “If we think about AI saving us 10 minutes a day, that equates to one week over a whole year, and so in an organisation like Shell with so many employees, that is actually quite a big saving.”

Representatives from Heineken and MMR Research Worldwide discussed whether AI personas, which are chat-based interfaces built from consumer data, could generate novel and diverse product ideas. The organisations worked on such a project, which also introduced an “agentic ideation” feature which enriched ideas using internal and external data.

Sophie Read, innovation insight manager at Heineken, said: “The ‘agentic ideation’ pipeline has taken things further. It’s shown us how we can use all the data we have in a much more meaningful and powerful way.

“The outputs were measurably more novel. These ideas were not just new and different, they resonated with a specific segment of consumers actively looking for something new.”

A common theme throughout the day was how AI was a research enabler, not a replacement, which was highlighted by an adviser to investment firm JAB Pet Services, who worked on a project with Inspirient, unlocking its quantitative data so it was accessible to everyone in the organisation, not just data analysts.

Mark Barraud, special adviser at JAB Pet Services, said: “The insights team is not out of business. That judgment, that critical thinking, especially at the front, is super important.”

Others agreed that AI would not lead to a decimation of market research jobs, but instead was an opportunity for researchers to learn new skills.

Meanwhile, Comic Relief, which has an insight team of just one person and a one-person innovation team, explained how it was using AI to fire up its research offering.  The charity has been working with AI video research specialist Motives, testing out AI-moderated video interviews.

Robert Alleyne, senior audience strategist at Comic Relief, highlighted the benefit of AI video interviews.

He said: “They are insights that would never have been shared with me if I was doing the interview. That part of Motive is really powerful.”

Alleyne said the charity had also been using AI to preview audience reaction to comedy sketches ahead of the sketches going live on Red Nose day.

Insight leaders from Co-op, like many others, talked about the efficiencies they were gaining through AI.

Working with Human8, the retailer used AI to transform its store experience, blending AI-driven insights with “human empathy”.

Molly Hilton, research analyst at Co-op, said: “We had all that data that we were sitting on, that would have taken us traditionally days, weeks, maybe even months to manually mine through and analyse but with AI you can do it a lot quicker. We also save time by mini stress testing everything early on.”

Many of the researchers who spoke at the conference said they had had minimal pushback from colleagues after introducing AI-powered research initiatives.