Ampersand has closed a $15 million funding round, the company said. The startup builds technology that connects AI agents to legacy enterprise software — the old, deeply embedded systems that most large companies still run on. The investment is a bet that the next wave of AI adoption won't replace those systems outright but will need to plug into them.
That's a messier problem than it sounds. Legacy software — think decades-old ERP platforms, mainframes, and custom databases — wasn't designed for AI agents to query, interpret, or act on. Ampersand's product sits in between, translating between modern AI models and the systems businesses already rely on.
Why investors are betting on the middleware layer
The funding reflects a broader shift in how companies are approaching AI. For the past few years, the conversation has centered on building new AI-native tools from scratch. But the reality inside most enterprises is different: critical data and workflows live in systems that predate the cloud, let alone large language models.
Ripping those out isn't realistic. They're too central to operations, too expensive to replace, and too risky to migrate on a whim. So the market has started to reward companies that make the old stack talk to the new one. Ampersand is one of a growing number of startups positioning themselves as the connective tissue between AI agents and the enterprise back end.
The potential upside is significant. If AI agents can read from and write to legacy systems reliably, they can automate tasks that currently require human operators to shuttle data between applications. That could reshape how businesses handle everything from supply chain management to customer records. But the technical bar is high — legacy systems are notoriously brittle, poorly documented, and varied across companies.
The integration challenge
Connecting AI to legacy software isn't just a technical problem. It's an operational one. Enterprises have spent years building security protocols, compliance checks, and access controls around their existing systems. Any tool that touches those systems has to satisfy the same requirements — and then some, since AI agents introduce new risks around data leakage and unintended actions.
Ampersand hasn't disclosed specific customers or deployment details. What's clear from the funding is that investors see enough demand to back the approach. The $15 million gives the company room to build out its product and prove the integration can work at scale.
The company hasn't said exactly how it will spend the money or when its product will be generally available. The broader question is whether AI-legacy integration becomes a standalone category or gets absorbed into larger enterprise software platforms. For now, Ampersand is betting that the gap between AI agents and old systems is wide enough — and painful enough — to support a dedicated business.
