Sandra Buckingham
When Innovation City member sens began five years ago, its founders were already using AI-driven predictive analytics to rethink how businesses understand consumers. Since then, the technology has undergone an extraordinary transformation - and so has the business.
Five years later, sens is introducing Conversation Monitor, its new social listening product that turns public customer conversations into actionable business insights. Co-founders Per Lagerstrom and Jacqui Young reflect on what has changed, what they have learned and what comes next.
Q: Take us back five years. What were you using AI to do, and what convinced you to build sens around it?
We wanted to understand people at scale without losing the richness of their conversations. We also wanted to fuse quantitative and qualitative research, fundamentally reimagining the economics of deep insight research so that more businesses could access it.
At the time, we described the work more as predictive analytics than AI. The ambition was to make a depth of understanding that had been expensive and difficult to obtain more accessible. That is why we built sens around the technology.
Q: How has the business changed as AI has improved and become widely accessible? Was there a big pivot, or a series of smaller reinventions?
There have been many smaller reinventions, but the past year has brought a fundamental change. More capable models from Anthropic, combined with the introduction of agents, have allowed us to pursue something that previously felt more like an aspiration: making sens truly AI-native.
The first step is for specific processes to run without human intervention. Eventually, our ambition is for the whole business to operate autonomously. That means rethinking how work happens across sens, rather than simply using AI to help someone perform an individual task.
Q: Per, you often write about moving beyond AI tools to changing how a business operates. What have you had to rethink or redesign inside sens itself?
Per: We have had to examine the whole journey. But I would flip the way we talk about it: the transformation comes from what AI allows us to do differently.
Increasingly, language models can run processes through instructions and prompts, rather than us having to build everything around static workflows. That changes how we organise and execute the work.
Q: Jacqui, what have five years of customer conversations taught you about the difference between what people say and what they mean? What has surprised you most?
Jacqui: What people say and what they mean are often two very different things. Understanding that difference is a signal interpretation problem. Much of our work involves cleaning up those signals and reading beneath the surface to understand the underlying intent.
If we had to identify the central capability of sens, it would be modelling consumer intent. We unpack that intent into its drivers, then unpack those drivers into what a business can do about them.
That is also why the lessons in our fifth-birthday reflection piece, Five Years In, We’re Still Listening, go beyond taking customer responses at face value. You need to understand what is driving them.
Q: As the technology has become more accessible, what has become more valuable about the expertise sens brings? Where does human judgement still make the difference?
Per: Human judgement still makes a huge difference. AI can ask questions, but asking the right questions in the right way to obtain an unbiased response requires expertise. So does analysing the resulting data properly.
There is substantial intellectual property in that work which does not simply arrive with access to a model. It is relevant to what I explore in my article The Human Moat Is Knowing What Matters, and Connecting the Dots: knowing what to ask, how to interpret the evidence and which connections matter.
Q: What client need or frustration led to Conversation Monitor, and how does it build on what you’ve learnt over these five years?
The market for research is changing. There is still a need for episodic research: a client needs information to inform a particular decision. But clients increasingly want a continuous understanding of what is happening in their relationship with consumers.
Conversation Monitor forms part of that, alongside our measures of Love, Loyalty, Trust and X-Factor, and the ability to conduct rapid, specific deep dives when something needs further investigation.
It brings together public conversations across reviews, app stores, forums, complaint sites and social platforms, separates meaningful signals from noise and connects findings to business categories and service touchpoints.
Together, those capabilities allow us to follow what consumers think and feel over time, investigate changes and help clients decide how to respond.
Q: Can you give us a practical example of how it turns a customer conversation into a business decision?
An illustrative example would be a business whose customers repeatedly describe its service as “a mission”. A sentiment score might register dissatisfaction, but the business needs to understand what is creating the effort.
Perhaps customers have to repeat information every time they contact support, or struggle to get clarity about refunds. If the evidence consistently points to one touchpoint, there is a specific process to investigate and improve. A focused deep dive can help establish the underlying drivers, and subsequent monitoring can show whether the frustration persists.
The value comes from connecting what customers are saying to a decision the business can act on.
Q: What did you get wrong along the way, and what would you do differently knowing what you know now?
With hindsight, I would have started the move towards continuous understanding alongside episodic research earlier. Some clients were ready, and we did not read that signal quickly enough. We were probably about six months later than we could have been.
The other major development over the past 18 months has been the quality of our recommendation engines. Recommendations to clients are now generated by what we call the sens brain, which has developed a strong ability to help clients understand what needs to be done. Improving that capability has been a continuous journey.
Q: What advice would you give fellow Innovation City members trying to build or adapt their businesses with AI?
Don’t make the business about AI. Calling something AI can sound like a hammer looking for a nail, which understates its transformative potential. Use it to improve effectiveness, efficiency and the experience you deliver to clients and consumers.
The second point is to understand the difference between agents and the underlying models. They are not the same. An agent can interact with multiple models, and that layer is critical to understanding what you can build.
As a founder, you need to spend time mastering both. Start with what you want to improve, then understand how the technology can help you achieve it.
Q: What’s next for sens and what problem are you most interested in solving now?
Our challenge now is to demonstrate that we can scale this.
We are focused on building an enterprise solution that brings together Conversation Monitor, our LLTX (Love, Loyalty, Trust and X-Factor) measures, and deep-dive capabilities in a dashboard that CMOs can integrate into their marketing operations. That is our single-minded focus at this point.