Leading Agave's Series A: AI Infrastructure for the Construction Industry

Some of the most interesting AI companies right now are working in the most vexing conditions — places where the data is unique, the systems are old, and the room for error is small. Construction gladly delivers on all of that difficulty and more.

Construction is not code (yet). It is messy and multi-party. It is trapped inside incumbent software written decades ago, much of it on-prem, almost none of it built for the modern world. A single project can run for years, change scope midstream, and pull dozens of unrelated companies into the same ledger. Every job is its own business. Margins live or die on whether someone, somewhere, reconciled the numbers correctly.

That’s a brutal environment for AI. Rather than simply producing something plausible, a construction agent has to reason about software it didn't write, understand intent that is rarely stated cleanly, and navigate to an outcome that is verifiable — because in construction, close enough is not an answer.

These are the markets where lessons are hard-fought, trust is earned through consistency, and value accrues to the patient.

The team at Agave — Tom, John, and Pooria — spent four years doing the unglamorous work. They interfaced with the brittle mesh of ERP and project-management systems that sit beneath the category, and built proprietary, two-way connections into more than fourteen of them. Read-only tools can look. Agave can act.

That distinction is everything. Most AI products are guests inside someone else's system, hoping the data they need happens to be there. Agave designed and owns the rails its agents run on: a proprietary infrastructure built to carry a workflow end to end.

We are entering an infrastructure build cycle that will rattle the foundations of how construction works. And it arrives at a moment when the industry can’t look away: 41% of the construction workforce will retire in the next five years, and barely 10% is under 25. The people who hold the institutional knowledge are walking out the door faster than anyone is walking in. Agave turns that knowledge into workflows that don't retire.

It’s proven out in the durable business they’ve built. Agave has been profitable for over two years. Revenue is tripling year over year. It is the fastest-growing agent in its vertical, already running across more than 80,000 projects and over $100 billion in project volume, and saving customers dozens of hours every month.

Most of the energy in AI today goes into building new software. The larger prize is teaching AI to operate the software the world already runs on — to get more out of what companies already depend on. Agave is doing that in one of the hardest corners of the real economy, and winning.

Fittingly, Agave is named for a plant that thrives in hard ground. We led their seed round, and look forward to continuing this partnership by leading their Series A.