The companies making real progress with AI aren't experimenting. They're picking targets, setting deadlines, and holding the line.
At Trax, we said we wanted to reduce certain processes by 70%. We said we wanted 100% of specific workflows running through AI models by a set date. And we told the teams we weren't budging. Pressure makes diamonds.
We Failed More Times Than I Can Count
Did we deliver everything on time, perfectly, every time? Not even close. We failed more times than I can count. There's a famous story about Edison. After thousands of failed experiments, someone asked him about the lack of results. He said he'd gotten plenty of results because he now knew several thousand things that won't work. I now know a lot of ways not to deploy AI.
But almost 100% of our paper and similar invoices are now running through artificial intelligence. We've ditched OCR. We no longer go through long, complicated carrier onboardings. We're using AI to accelerate carrier onboarding. We have deployed agents to help clear exceptions. All of that is in production today.
When we first tried these things, they went poorly…really poorly. But we had a target and a deadline and we held everybody accountable to both. When something didn't work, we told the team congratulations, you just found a way it doesn't work. Dust yourself off and get back out there.
Nobody Goes Out on a Limb If You Yell at Them When They Fall
If you get upset with people when they fail and start yelling at them, nobody's going to take a risk for you. They're going to stay close to the tree trunk and hold on to safety. That's not what we're looking for. We want people who are willing to move forward and push into new territory. We call this “living on the short branches”. Again, if every time someone falls, they get yelled at, no one is going to leave the safety of the trunk.
Low Risk Experiments Produce Low Value Results
If you're taking a low-risk area of your business and just trying AI out, you're going to end up with a low-value solution that never gets deployed.
Make a big move. Go tackle a major problem. There's very little risk in doing that because you don't have to push it into production until it's working. Put a number out there. Tell the team this is what we're going to hit by this date. Humans are incredible in their capacity to innovate and perform. We underestimate each other daily.
This Is the Cloud Conversation All Over Again
I was in on-prem software just like everybody else my age. Everybody said nobody's moving infrastructure to the cloud. I remember going, wait…yeah, they are…100% THEY ARE. Absolutely. You could see it coming clear as day.
But the first conversation with a customer was always the same. “I don't know how secure the cloud is. We don't do that here. Everything's on-premises. We want to be in control.”
Now look at where we are with AI. Same conversation. Very complicated document from the legal department that says you really can't do anything with it. Exactly the same.
Five Years From Now This Will Be an Afterthought
Five years (or less!) from now, a lot of people are going to be very comfortable with AI. It's going to be an afterthought, not something we're all trying to figure out.
If you're still in experiment mode even two years from now, you're way behind. I'd argue you're way behind right now. The companies that are setting targets, holding teams accountable and creating space for failure are the ones that are going to come out ahead.
Wondering where to start with deploying AI across your supply chain? Explore our resources or connect with the Trax team to get the conversation going.