Delivering high-value insights and strategy using AI: business results from AI

So fiona lets turn to you for another question that i know a lot of companies struggle with. They have figured out how to do pilots with ai or proofs of concept or demonstration projects, theyre able to show the value, and then they struggle with how to adapt that and scale it enterprise-wide. So any insights that you have or advice for others on how to achieve ai enterprise-wide at scale um. Certainly, i think one of the the key

Ways that weve weve worked at it at wayfair is to really tightly connect the work that were doing in a ml with the business right. So a lot of it is going to be part of the business function and understanding what the business needs are and and the customer needs obviously as well and you’re tightly coupled to that. So its not as though the aiml team

Is building something that that is sort of incongruent with everything else that the business and the customer wants? So in order to do that, it has to be that very tight alignment, i would say also in the planning process. Uh, you want to make sure you bring in the the data science leaders early on, because it could help in terms of how you id what it is that you’re trying to build from a product perspective and how you can leverage as much as possible right um Data science and ai and ml and its

Having those leaders in the room having a seat at that table where you can really come up with some innovative ways in which you can leverage the technology stack, we have um been move, weve, been moving more and more towards more of a federated model. So again, the reporting structures and how the orgs are aligned are really more functional sort of business within business areas, but you still have

Some amount of um sort of, i would say, like a machine learning or data science guild that ensures that there is the ability to build out some of the machine, learning platforms and the services and models that can be leveraged across different areas. Right and probably the last thing, i would say in terms of how weve been

Able to scale too is we do also have a um, a smaller organization called wavefriendx and theyre really focused on experimentation and looking at so a lot of the the things i talked about imagery, creating 3d models and a lot of the ar work that were doing Um, they would probably pilot first but then very quickly, um they work with the production with the engineering and data science teams that are within the rest of the wafer organization to to push that into production. So there’s, not a huge long.

Cycle between whats experimented on and what goes into production, so there’s a fairly tight loop, but its you know. We found it useful to also have a a small innovation team that can experiment, try out new technologies and vet out what makes sense what doesnt make sense and then we can consume that faucet and roll it out um into kind of like a higher skilled way.