One of our customers recently had the opportunity to close a large nationwide account. Real new revenue, the kind of deal that changes a year for a customer.
There was a catch. That account had an AI project running internally, and it changed what they needed from a supplier. Our customer called us in to talk it through with them.
Fifteen minutes into that call, our customer was able to say yes and the customer integrated in a day.
I want to walk through what happened, because I think the interesting part isn't the technology. It's what it meant for our customer to be the supplier who could say yes.
The nationwide account wasn't getting rid of punchout. Once an order is approved in their purchasing system, it still flows through the punchout integration the way it always has. That rail works, procurement trusts it, and there was no reason to touch it.
What they were rebuilding was the shopping experience. Rather than send a buyer out to a supplier's commerce site to browse, they wanted their AI to query suppliers directly. What's available, what are the attributes, run a search, pull current pricing. Do the specific commerce work up front, then hand the approved order back to punchout.
That's an important distinction, and I think it's where much of this is headed. The transaction layer stays where it is. The layer in front of it is the part being rebuilt. A supplier who can't feed that layer doesn't lose the order at checkout. They just aren't there while the buyer is shopping.
A lot of suppliers in that position would have had to say "let us scope that." Some would have had to say no. Even the ones who could eventually get there would be looking at a project, a timeline, and a conversation with their software vendor about whether it's on the roadmap.
Our customer said yes on the call. It's all out of the box. Here's where you go to get your API key. Here's the API set.
Nobody wrote a custom integration. Nobody opened a project plan. Nobody called us afterward to ask for a feature. Their customer had what they needed and was integrated the next day.
Here's why that was possible, and it's the part I'd want any distributor to understand.
When a customer signs up with Kodaris, they instantly become a tech company offering open REST APIs direct to their own customers. Their customers can generate their own API keys and access their own data with that key, locked entirely to their own data. It's branded with our customer's logo, their usage terms, their relationship.
They didn't build any of it. They don't maintain any of it. They didn't ask for it, and until that call, most of them had no particular reason to think about it. It was just there.
For a mid-market distributor, offering that kind of thing used to mean an internal engineering team and a multi-year project. Now it's part of what they bought.
The foundation underneath is not the story here. We built Kodaris as a platform from the ground up, with open, documented, composable REST APIs, and we used those same APIs to build our own applications. That decision was made years ago, long before AI was a factor. It just happens that AI works very well with composable REST APIs, so work that was already done turned into something our customers can hand to their customers today.
That call wasn't a one-off. We have customers' customers who have built their own CRM-style applications on top of Kodaris, using the APIs and AI together. They're not asking anybody to build features for them. They're building against the platform themselves, at their own pace, for their own workflows.
Different ask, same answer. Yes, go ahead.
The large providers are all moving this direction. Salesforce, AWS, ServiceNow are putting out APIs and tooling for AI, and that's the right call. It tells you where the industry is headed. The question for distribution and manufacturing is whether the systems we actually run every day get there too, and how long we have to wait.
Most AI evaluations I see are organized around a specific use case. Can it do product search. Can it write descriptions? Can it answer customer questions? Those are fair questions, but they have short shelf lives. AI is turning over fast enough that whatever wins the feature comparison today probably isn't the differentiator two years from now.
Now let's look down the road three to five years. The question that holds up over that stretch is a different one. How does your platform partner support your business and your customers, end to end? When your customer shows up with something you didn't plan for, can you say yes? Or does the answer depend on a roadmap you don't control?
That's not a feature comparison. That's whether the platform you picked is a strategic advantage or a ceiling.
Our customer didn't win that account because of an API. Nobody buys from a distributor because of an API.
They won it because when their customer asked for something hard, on a live call,, they had an answer that day. Not "we'll look into it." Yes, today, here's how.
You don't get many of those moments, and they're the ones people remember you for. That's what a platform partner is supposed to be for.