Why I Built Revver Contract Intelligence

This past summer I killed 4 months of my own work. Dragged it to the trash. Months of contract templates built with people who’ve got decades of procurement experience. And it was the 2nd time in a year I’d done that.

Both times, the thing I killed worked. I could’ve shipped either version and most people would never have known the difference. But I run this company on one rule, quality over everything, and I’ve got to be proud of what this system produces. So twice I sat down, had a hard conversation with myself, and said the same thing. This isn’t it.

To explain why a guy with no engineering background was even in a position to delete months of his own code, I’ve got to back up.

The Plane

I’m not a programmer. For most of my life I was a great user of tools and nowhere near tech savvy enough to build them, and that was the wall between me and everything I wanted to make. Then AI coding tools showed up, and the wall crumbled.

About 3 months into building the first version, honestly a garbage version, I was on a plane reading Rick Rubin’s The Creative Act, and a passage about childhood dreams sent me somewhere I hadn’t been in a long time. As a kid, I’d lose myself in drawing for hours. I had a drafting table and I designed buildings all day. I wanted to be an architect. Fast forward to high school, and an art teacher ridiculed me for not knowing color theory. I’m colorblind. Known it all my life. Not black, white, and grey, but enough to make poor outfit choices. At the time, I was embarrassed enough that I dropped art completely, got serious about math and science, and architecture went out the door with it. For close to 25 years.

Sitting on that plane, I whispered aloud to myself, “I may not be the type of architect I thought I’d be, but I can most certainly call myself an architect today.”

I wasn’t expecting how hard that would hit. It validated something I didn’t even know needed validating, and what it lit in me wasn’t relief. It was fuel. Hell yes. This is it. Let’s go.

These days I tell my kids my color blindness isn’t a bug, it’s a feature. I see the world differently than everybody else. That’s my advantage. They love that, and I believe every word of it.

The Product Was Walking Out the Door

Here’s what I was actually building on that plane, and where the idea came from.

I’ve spent my career in startups on the revenue side. At the last one, a staffing-tech company, I helped grow revenue from $4M to $22M in about 18 months. My niche was public accounting and consulting firms, teams at places like Grant Thornton, BDO, and RSM. Firms whose entire product is their collective expertise and judgment.

Those firms have mandatory retirement ages. And I kept watching the same movie. The most experienced partners would age out and leave, and everybody called it people leaving. It’s not. When your product is expertise, the product is leaving the building.

I couldn’t let that go. Could you capture what’s in an expert’s head, the judgment, the exceptions, the gut feelings built over decades, and structure it so the organization keeps it after the person’s gone? I started calling it knowledge architecture. An institutional memory that doesn’t retire.

I also knew one thing for certain. No billion-dollar CPA firm was going to let an unproven guy test that thesis on their business.

The Deal

I took the thesis somewhere it could actually be tested. A high-end boutique procurement advisory firm, two founders and several senior advisors with more than 100 years of combined procurement experience between them. I didn’t pitch them knowledge capture. I asked one question. What do you wish existed?

Then I made my offer. I’ll build it for free. I’ll cover every cost. It might be a total flop. But if it works, your firm uses it free for as long as your firm exists. In return, you’re my design partners, and you answer every question I’ve got.

What followed was months of interviews. Not procurement theory, real contracts, one after another, with the people who’d lived them. Procurement veterans, attorneys, risk and compliance practitioners, each reading the same agreement through a completely different lens. What matters in this clause and why. Where your gut goes first. What you check on instinct because you’ve seen it burn somebody before.

1,200 Hours

The first month of building is pure dopamine. You talk to a machine and it makes what you describe. Sunshine and rainbows. Then reality shows up.

Version 1 used a cheap AI model to read contract clauses and an expensive one to reason over what it found. Efficient on paper. Except nothing checked the links in the chain, and it doesn’t matter how brilliant your final reasoning is if it’s reasoning over bad data. So we killed it.

Version 2 put one highly capable model on the whole document, then filtered its analysis through a series of contract templates and benchmarks, the ones I’d spent 4 months building. What a good MSA looks like, a good SOW, a good NDA. The filtering felt rigorous. It was a cage. I had one of the strongest legal reasoning engines ever built and I was constraining it to the 20-25 things I’d told it to look for.

Worse, the cage exposed the deepest flaw, the most important thing I learned in the entire build. Tell a system to find what’s missing and it becomes purely additive. It’ll only ever add. It’ll never read a contract and say this clause doesn’t belong here, this is harmful, strike it. Real experts subtract. That realization changed the product, and it’s big enough that I’m writing about it separately.

Version 3 reads the entire contract first, ungated. No filter, no checklist, tell me everything you see. Only then does it reconcile its findings against a dense library of expert standards, judgment, and negotiating positions. That’s the system running today, roughly 1,200 hours and 8 months after the first line.

If there’s an upside to killing your own work twice, it’s this. I’ve got no sunk-cost bias, and I’ve proven it. I’m okay being wrong on the way to being right, and I’d rather be fast and honest than slow and precious.

What We Built Differently, On Purpose

Plenty of tools read contracts now. Here’s what I took special care to build, because each of these came straight out of those months of interviews, and I don’t believe you’ll find them together anywhere else.

It Reads the Whole Deal, Not Clauses

A contract isn’t a stack of independent provisions, it’s a system, and a clause can look fine until another clause changes what it means. An indemnity that reads strong on page 8 can be quietly gutted by a liability cap on page 19, or by an amendment signed 2 years later. Revver Contract Intelligence evaluates those interactions across the entire contract family, MSA, SOWs, amendments, all of it. Checklist tools structurally can’t do this, and it’s where the expensive problems hide.

It Knows How to Subtract

Because of the additive lesson, this system can flag provisions that don’t belong, or ones that are actively harmful, and draft the cut. With AI now writing a growing share of the world’s contracts, and AI drafting being additive by nature, I think subtraction is about to become the scarcest skill in contract review.

It Learns Your Judgment, and Your Judgment Stays Yours

Every correction, exception, and decision your team makes becomes part of your organization’s own knowledge, kept in readable prose you own. Your first contract gets Revver’s expert baseline. Contract 100 gets all that plus everything your organization taught it along the way. That’s the knowledge architecture thesis, finally built. The judgment stays.

And If You Leave, It All Leaves with You

I sold data for years, and I sat across from deals that made me feel gross. We decided on day one that everything a customer builds inside Revver, the decisions, the standards, the accumulated knowledge, walks out the door with them. No fee, no hostage negotiation. It’s not a growth strategy. It’s just right.

Why Should You Care

If you’re a founder, this is the stuff that decides whether you can get out of a contract the year everything changes. If you’re a CFO or a GC, it’s a 2nd set of expert eyes that never gets tired and never forgets a decision. If you run Procurement, it’s your best negotiator’s instincts, available to your whole team. And if you’re just someone who likes watching one person take a swing at an entrenched industry, stick around.

What Revver Contract Intelligence Solves

Everywhere I look, companies are asking people to do more with less. When I hear more work, I hear more time, and time is the most valuable thing any of us has. Hands down. I recently talked with the CFO of a 9-figure company who spends weekends reviewing technology contracts with his general counsel, because the workweek is already spoken for.

That’s who I built this for. Not to take the work off his plate. To do the review better than a tired Saturday afternoon ever could, and give him the Saturday back.

If you ever want to reach out, I’m at sp@revver.ai, or find me on LinkedIn.