Yoav Kutner is the Co-Founder and CEO of OroCommerce, a B2B-focused commerce platform for manufacturers, wholesalers and distributors.
A CIO at a $2 billion distributor is sitting on a board mandate to deploy AI. She also manages four different ERPs acquired through years of M&A, a green-screen order system from 2003 and a software vendor promising her that the next massive ERP upgrade will include AI natively.
So, she is waiting. This is the most expensive decision she will make this year, and she doesn’t know it yet.
Everyone knows artificial intelligence (AI) is the future of B2B commerce, but nobody knows how to plug an advanced large language model into a 20-year-old AS/400 system.
Waiting for the ERP vendor is the rational call. It’s also how two years disappear. According to IDC, 21.4% of companies plan to renew with their current ERP provider specifically because generative AI is included in the next release. By the time that upgrade ships, the AI features are often shallow and the window has closed.
A standard enterprise ERP replacement takes two to five years. Advanced AI capabilities are moving on six-month cycles. Pause for the migration, and your AI strategy is five generations obsolete before the project closes. This timeline mismatch leaves IT leaders feeling as if they have only two options.
They can tape a fragile AI chatbot onto the outside of their legacy stack, creating security risks and data hallucinations.
Or they conclude that the ERP itself is the problem, and commit to a full replacement—a project that absorbs IT resources for years and pushes every other initiative, including AI, to the back of the queue.
Neither works. And the data reflects the bind: in a recent survey of B2B leaders, 53% cited legacy system integration as their biggest barrier to AI adoption, while 67% admitted their data formats aren’t standardized across systems.
What Banks Know That Distributors Don’t
To solve this, B2B leaders need to look at how global banks handled their own legacy crisis.
Most major financial institutions today still run on COBOL, a programming language from 1959.
They didn’t rip it out because replacing a core banking ledger is incredibly dangerous. Instead, they built modern API layers, mobile apps and highly advanced AI fraud-detection systems on top of the mainframe.
Distributors and manufacturers must adopt the same mindset. Don’t rip out a functional ERP just because it’s old. Treat your ERP the way a bank treats COBOL: let it be the dumb, reliable calculator. Build a “smart skin” around it to handle the intelligence.
The Strangler Pattern
In software engineering, this is known as the Strangler Pattern, named after the strangler fig tree, which wraps itself around a host tree and gradually takes over its structural role without killing it.
Applied to B2B commerce, you wrap your legacy system in an intermediary layer instead of tearing it down. For B2B organizations, this means putting a unified digital commerce platform in front of your back-office systems.
You gradually migrate the dynamic, customer-facing business logic (like complex contract pricing, custom catalogs and corporate account hierarchies) out of the ERP and into the commerce platform.
For most B2B organizations carrying years of technical debt, this is the most practical path to making AI work.
An ERP is a system of record for your ledger; it was never designed to manage the digital buying journey, semantic search or conversational commerce. AI requires unified, clean data to function. Once the commerce platform has “strangled” the old systems, it becomes the clean, unified brain of your operation.
You point your AI tools at this layer, not your messy, fragmented ERPs. Because the commerce platform already holds your business rules, the AI inherits your strict permissions automatically. It knows exactly which buyer is logged in, what their credit limit is and what their negotiated price should be.
Technology should serve the business, not the other way around. Ripping out a working supply chain system just to run an AI prompt is bad engineering and a worse business strategy.
Modernize the layer the customer touches, protect your margins and let your legacy infrastructure quietly do its job in the background.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
