AI assistants answer. Agents do the work.
Editor's note: This article is an automated speech-to-text transcription, edited lightly for clarity. For the full session, please watch the recording.
On June 11, 2026, Voxpopme gave a presentation comparing AI assistants and agents during the Quirk’s Virtual Sessions – Ensuring Data Quality, Security and Ethics series. Voxpopme was one of three organizations that participated in this series.
Tom Higgins, senior growth and product manager at Voxpopme, went through a study he conducted on consumer attitudes toward ultra processed and minimally processed foods. He wanted to see how the use of an LLM compared to an agent specifically designed to be used by market researchers. So, Higgins used ChatGPT and Voxpopme Compass for the comparison.
In this session, Higgins will illustrate the difference between LLMs and AI Agents through the data he collected in this study.
Session transcript
Joe Rydholm
Hello everybody. Welcome to our presentation, “AI assistants answer. Agents do the work.” I'm Quirk's Editor, Joe Rydholm, and before we get started, I just wanted to quickly go over the ways you can participate in today's discussion.
You can use the chat tab to interact with other attendees during the discussion, and you can use a Q&A tab to submit questions to the presenters. We'll address them in a live Q&A after the recording. Our session today is presented by Voxpopme. Enjoy!
Tom Higgins
Good afternoon, everybody and thank you so much for being here with us today. It's always an absolute pleasure to be here with the insights community. And hopefully we've got something fun for you today off the back of the launch of Voxpopme Compass, Voxpopme's very own AI agent for market research.
We wanted to test AI agents versus LLMs. So, we have something of a machine vs. machine research showdown for you today. Just so when you're making decisions as to what the right tools and methodologies are for your research and this kind of new AI and agentic era and all the other AI buzzwords that are floating around these days, you can make a decision grounded in evidence.
So, without further ado, I'll jump straight into it.
For those of you that don't know me, I'm Tom Higgins, Voxpopme’s senior growth and product marketing manager. I would love to connect with you all on LinkedIn. So, I'll share my LinkedIn link in the chat. Please reach out to me if you've got any questions. I think we've got time for Q&A at the end as well. So, I'll tackle some there too.
But without further ado, let's get into it.
So, the big question is like now that pretty much anyone can paste transcripts into ChatGPT, it's very reasonable as the researcher or insights professional to ask, “If a general model can read my study and summarize it, do I still need market research platforms?” There's a huge ecosystem of market research platforms out there. Yeah, “Do I still need those tools to conduct, understand research and share deliverables with my business?”
And today we're going to explore that and explore that through the showdown that we've created between LLMs and AI agents that are custom built for market research. How do we go about that?
Firstly, the study itself was the same study, with the same questions across both systems. I used Voxpopme Compass, Voxpopme’s AI agent to do one layer of analysis on a study about ultra processed versus minimally processed foods and consumer attitudes towards those two types of foods. As you can imagine, very, very divisive anyway. And essentially gave identical prompts to both Compass and a general purpose LLM. It was ChatGPT for anyone that's interested because I wanted to compare and contrast how the two performed and the results were really enlightening. So, we'll unpack them here.
So, firstly, before I could ask a single question to either the LLM or Compass I had to think about, “well, what's my source data for this study?”
And I guess the first barometer of ‘should I have a research platform’ is how am I actually going to conduct the research in the first place?
On the left-hand side without Compass and looking at going straight to an LLM, I couldn't just jump in there and be like, "Hey, what do you think about consumer attitudes to minimally and ultra processed foods?" Because I had no source data, what it would do in that instance is just go to general market data and lose the context of the custom piece of research that I'd run.
So, if you were to use Voxpopme Compass or an equivalent platform, you'd be out there, you've still got to go out and conduct the research. And in this case, it was qualitative in nature. So you find a way to conduct that video-based research. Then even just some of the baseline processes that happen when video-based research lands in a platform like Voxpopme, just having the transcription on every response, that needs to exist in order to give it to the LLM because it can't right now ingest all the video content that you would collect in that kind of study.
Then the final step I had to actually export the data into CSV format so I could give it to the LLM and begin that process.
So, right off the bat, your conversation with an LLM is delayed while you are out there conducting the research with Compass. It immediately had access to the study that I'd run in Voxpopme around this. And had I not had that, Compass would actually have the tools at its disposal to go and run that study as well if the data didn't preexist.