
Giving customers a conversation with their data.
Context. Back in 2023, I started talking about how AI could improve our reporting: surfacing the most relevant data and turning it into tangible actions and next steps. In design, we'd been focusing on storytelling in our dashboards, and had quickly realised that truly bringing the data alive would take more technology. What nobody said at the time was how hard it would be to deliver insights that were genuinely useful and relevant to our customers.
PG Forsta gathers survey data, consolidates it and produces dynamic reporting for its users. We do this across many industries, but by far the largest and most complex is US healthcare, where we hold a significant share of all the data from the last 40 years. Unprecedented data and knowledge, perfect for training AI models. Or so you'd think.
The opportunity . Developing the concept with key customers showed us how bespoke their processes were. Each of them did a lot with our data, in its own way, so simply replacing that with a model was going to take time.
What we did know was that people wanted more from the data. They wanted to interrogate it, see correlations, and ultimately be advised on where to spend their money to improve the product, service or experience they were responsible for. They wanted a conversation with their data. So that's what I set out to build.
Earning trust. Our first ideas focused on simple queries, generated dynamically by the system and based on the kind of work our consultancy team did. Working closely with them and key customers, we gradually built a model that grew in accuracy and usefulness, often through trial and error. Having the data didn't mean we could get the personalisation and insight right for every customer first time.
We started with a component inside existing dashboards that read the data on the page, suggested questions and encouraged users to engage. We quickly learnt that people wanted to drill deeper, challenging and questioning everything. They wouldn't take a recommendation from the system the way they would from one of our consultants. Trust had to be earned.
[Image: in-dashboard component]
A new workspace. So I moved us to a separate interface that behaved more like a chat window, giving customers far more room. Very quickly, it became their workspace, so much so that everyone on the trial wanted to convert to the product. They started choosing it over the traditional reports and dashboards, going straight to chat to ask their questions and generating reports and content to take back to their teams or bosses.
[Image: chat workspace]
Impact - It's changing how we do business. For years, our reporting meant static dashboards full of graphs, bar charts and tables, usually supplemented by a PowerPoint for customers to use internally. We'd been told it would be really hard to change people's mindsets. Customers moving to chat showed otherwise.
We got to market faster. Design prototypes, iterated with customers and tested quickly, meant first releases of new ideas shipped within a quarter, where previously they'd have taken several. Work that used to take weeks or months, designers now turned around in a few days.
We changed how teams work. Design and engineering now come together very early, building co-created backlogs and talking daily about barriers, so the work flows freely. As the AI tools improved, those conversations evolved with them, as we learned what customers would use, what they wanted and how quickly we could respond.
Today, AI powers comment summaries in dashboards, chat with data, chat with an expert, and patient coaches, all meeting our customers' changing needs.
What's next. We've also brought AI to the data-gathering side of the business. In surveys, it helps create more dynamic, tailored and targeted questions, improving the data set itself. In time, that means a more connected, human engagement with the people we survey, and a better human experience across our products.
Reflection. Design, and my job, is often about spotting new opportunities, telling the story, then developing the proposition with the business and customers to bring it to life. I set out to use AI to improve our products, but along the way I learnt so much more about our customers. New technology still has to earn trust and confidence if it's going to flourish and replace old ways of working.
