Hey everyone, I wanted to share an exciting update about a new service we’re offering at Sourcing Playground. It’s been quite a journey to get here, and I’m thrilled to tell you about how we’ve evolved and what we’re doing now with AI for procurement.
For years, we’ve been working with retail companies, helping them make multi-million dollar procurement and sourcing decisions. Our bread and butter has been providing market intelligence and trusted information to guide these choices. We’ve dealt with massive datasets, from certification and ESG data to trade information with millions of data points updated monthly.
As you can imagine, handling this much data came with its fair share of challenges. We faced issues with duplication, performance, accuracy, and making all this information actually usable for our clients. We knew we needed to find a way to make it easier to dig into and understand this wealth of data.
When AI and modern LLMs (Large Language Models) first come onto the scene, we saw the potential to tackle some of these challenges head-on. We’ve experimented with pretty much every major AI model out there – ChatGPT, Claude, Bard, Mistra, Llama – you name it, we’ve tried it.
Our experiments took us down some interesting paths. We started by testing if AI could be better at finding specific information than us and our users. Could we ask it to find “a clothing factory in Portugal” that matched certain criteria? We quickly learned that a more guided approach was necessary to prevent the AI from going off track or generating inaccurate information. Often it would bug out and provide false positives.
This led us to implement a more guided chat-based input, allowing users to express their needs in natural language while we provided structure to keep the AI focused. We later explored using AI for data cleaning, automated research, advanced curated insights and a few more things – all persistent headaches in our field.
Through all of this testing, we’ve learned a lot:
- Each AI model has its own strengths.
- The technology is progressing at a breakneck pace – we’re seeing new developments almost daily.
- AI isn’t a silver bullet. It has clear limitations and shouldn’t be treated as a cure-all.
- Some use cases are much better suited for AI than others. In some instances, better data processing is actually the answer.
- Make current data more accessible to procurement teams.
- Provide the answers procurement teams need to do their jobs effectively.
- Uncover hidden insights that could drive better decision-making.
- Use the latest technologies for competitive advantage
“Working with Sourcing Playground you always get individual attention and solutions specifically tailored to meet your exact needs. It’s enabled more impactful conversations with our supply base, focusing our limited resources on needle-moving initiatives. You work directly with their experts to build tools and solutions together, which is extremely refreshing after spending many years dealing with other service providers’ sales teams and canned reporting.”We’re really excited about the value it can bring to procurement teams. If you’re interested in learning more about how we might be able to help your company leverage AI for procurement insights, we’d love to hear from you. Heather Learn MoreSenior Director of Global Costing @ Under Armour