We talked endlessly about data and AI at DPW New York 2025. A universal truth is that the successful implementation of AI requires clean data. It doesn’t have to be perfect, but businesses certainly need to have a decent handle on their data before adopting AI tools successfully.
To help make this a reality, North American data and software company TealBook has recently announced a legal entity-based data model. It’s designed to resolve supplier records to the correct legal entities, map parent-child relationships, and enrich profiles with verifiable attributes, enabling accurate supplier data to flow seamlessly into procurement systems and AI applications. “This is part of a 12-year journey for TealBook,” says Stephany Lapierre, the company’s Founder and CEO. “Our vision has always been to build a way to enable procurement organisations to have high quality data with a lot of integrity. That way, you give them the trust they need to put data directly into their systems.
“Twelve years ago, we underestimated the complexity of getting large enterprises to trust a third-party data solution. As part of our journey, we started using AI early on to find information where it exists on supplier websites and databases. We also started creating digital profiles in a structured way for procurement to access it, match it to their vendor master, and use it.”
TealBook’s data evolution
But, again, at the beginning, TealBook couldn’t be sure whether the data was high enough quality. In 2017, the company was primarily known as a supplier discovery application. It was positioned as a pre-sourcing engine to help procurement teams identify alternative suppliers. At the time, TealBook’s data and models enabled it to determine which companies were similar to others. This meant users could search and find comparable suppliers to expand their sourcing options.
“But that was just a way for us to deliver something that was underserved in the market,” Lapierre continues. “Then our customers started asking for certificates, which are hard to collect and match. They needed cleaner data. They felt they were under-reporting. So in 2018, we started to see whether our technology could refine the data more. We focused on certificates and supplier diversity. We collected great use cases along this journey, and the vision never wavered.
“Just last year we released a new technology – completely different, really sophisticated – allowing us to pull from a lot more data sources. We have provenance so our customers can actually verify where the data’s coming from. We can match it to vendor masters. And now, we also have this new model that includes 230 million verifiable global legal entities from across 145 countries’ registries. We marry this with global parent and child hierarchy, which is really hard for our customers to match themselves.”
Partnership with Kraft Heinz
Now, after 12 years of that vision, TealBook is deeply proud of what it’s achieved. Part of its ability to get to this point is due to early adoption from key customers. Kraft Heinz is a business which Lapierre describes as a “co-innovation partner”, and has been invaluable in helping TealBook achieve its recent goals.
From the perspective of Stefanie Fink, Head of Global Data and Digital Procurement at Kraft Heinz, the partnership has been an immediately valuable one. “It really started with having a visionary, like-minded relationship,” she says. “That’s an important piece of it, because my vision for procurement is that we are partners in our enterprise.
“In order for us to do our jobs, we have to bring in the right data for use. This is where Stephany’s partnership and vision really resonated. We were really looking for diversity and we could make things easier for our partners, while making sure we had the right people in our ecosystem. We also had to lift up the hood and see what was underneath everything we’ve got. Stephany brought our vision to life. TealBook has evolved too, as we’ve seen; it’s more about orchestration and software-as-a-service. It has been a partnership of need and we cannot continue to do other things without this kind of partnership around data.”
When initially dabbling with this relationship, Fink was clear that Kraft Heinz had no desire to be taking care of more stuff. What she wanted from TealBook was a strong focus on good quality data. After last year’s product release from TealBook, Kraft Heinz already saw its data enriched by 25%. The recently-announced new data model gives the business and TealBook’s other customers the right structure tied to a legal entity, which is a highly credible anchor. “We’re able to do entity resolution – all automated – remove all the duplicates, and then you start with a clean, digitised vendor master,” says Lapierre. “That’s what brings further enrichment.”
The challenge of assessing data quality
Assessing its data before involving TealBook was important for Kraft Heinz, but challenging for such a large organisation. “We had to fail first and fail fast,” says Fink. “We tried some AI around fixing things early, but that didn’t work for us. It was a real eye-opener, realising where this next evolution could take us. Particularly regarding focusing on AI and agents for the right things, not the meaningless things. Before, we were asking agents to tell us if things were duplicates, when we should have been asking: what do these suppliers offer? Where is the innovation? Where is the value?”
What surprised Fink most when looking under Kraft Heinz’s hood was the lack of attention that was being paid to what the business was doing. “It was amazing that nobody had questioned it sooner,” she says. “So I said, let’s take this as a crawl, walk, run approach. I have a wonderful CPO who really understands where we want procurement to go as a function. She was excited about us just getting it done and getting people involved, and that’s what it takes: real pride in ownership of the data.”
Getting engrossed in GenAI
True partnership and an all-in approach has enabled Kraft Heinz to work successfully with AI. This is something some businesses are struggling with as the conversation around artificial intelligence grows louder. For Lapierre, as the CEO of a tech company, adopting AI successfully has meant trying and failing and being fully entrenched in AI as it has evolved.
“We’ve been using AI in our technology since 2016,” she states. “We’re an early adopter. We’d be talking about scraping data, and data in the cloud, and AI models, and our customers’ pupils would widen in surprise. We’ve come a long way and the market has come a long way.
“The technology we deliver today wouldn’t be possible without the AI tools now at our disposal. We used to build models; we don’t do that anymore. We spend a lot of time investing in engineers to build and test models. That’s made us so much more efficient. I use GenAI every day for so many things now. I’m encouraging my team to be so involved in AI. That’s how you build expertise. You need really strong expertise to use GenAI well.
“Getting good with AI is about taking risks and having a leadership team that pushes for new things. Suddenly, the successful use of AI becomes a habit.”