A few months ago, we released a tool that analyzes a company’s global regulatory exposure in about sixty seconds. Relevant frameworks. Jurisdictional overlap. Upcoming deadlines. A risk score.
We made it free. No gating. No lead form. No sales call.
Several people asked me why. The answer matters beyond our company, because a lot of AI-enabled enterprise software businesses are about to face the same decision — and most of them are going to get it wrong.
What AI Is Actually Changing
The common framing is that AI replaces expert work. That’s the shallow version.
The deeper version is that AI is moving the line between what parts of expert work are scarce and what parts are abundant. Expertise isn’t obsolete. Certain outputs of expertise are just becoming dramatically cheaper to produce.
In compliance, the work has always had two layers.
One layer is pattern recognition. Reading regulations. Mapping them to a company’s operating footprint. Identifying frameworks. Flagging deadlines.
The other layer is judgment under ambiguity. How a regulation applies to a specific supplier. Whether evidence is sufficient. What the intent of a requirement is when the letter is unclear. Which edge cases matter.
AI now does the first layer quickly and reliably. It has not meaningfully touched the second. And for the foreseeable future, it won’t.
That split is what most AI strategy conversations miss. It’s not “AI replaces experts.” It’s not “AI augments experts.” It’s more specific than that. AI is commoditizing the scaffolding layer of expert work. The judgment layer is untouched.
Why Charging for the Snapshot Is a Bad Business
That distinction shaped how we thought about the tool.
The diagnostic it produces is the scaffolding layer. It’s the part of a compliance assessment that used to take weeks and cost thousands of dollars. Not because the work was rare. Because the labor to do it carefully was.
That labor isn’t scarce anymore.
Within a year, every serious vendor in this space will have something similar. The technology isn’t proprietary. The prompts aren’t defensible. Charging for the snapshot today means building revenue on something the market is about to decide is a commodity.
Two options when you see that coming.
Collect whatever revenue you can before compression sets in. Or use the fact that you can produce the snapshot cheaply to accomplish something more useful.
We chose the second. Three reasons.
First, it helps buyers make the internal case. The hardest problem compliance leaders face right now isn’t buying software. It’s explaining to their executive team why the current approach isn’t working. They know something is off. They need a clean artifact that makes the exposure legible to a CFO. A sixty-second assessment, with no sales strings, does that better than anything else we could give them.
Second, it inverts the trust posture. AI-era enterprise software has a trust problem. Every vendor is pitching transformation. Buyers are appropriately skeptical. Giving something real away — no form, no call — tells buyers we’re confident in what sits behind it. They verify the output against their own situation. If they see something that warrants a conversation, they come to us. Different footing. Better conversations.
Third, it reflects where our value actually lives. The snapshot isn’t what we sell. What we sell is the system that operates continuously after the snapshot. Supplier evidence validation. Real-time regulatory monitoring. Mapping obligations to specific products as regulations like PFAS restrictions and the EU’s Carbon Border Adjustment Mechanism evolve. That’s the judgment layer, operationalized. Months of engineering. Years of domain accumulation.
The snapshot is a preview of a state. The system is the state.
If our product were the snapshot, giving it away would be ruinous. It isn’t. So giving it away is the most honest way to describe what we actually do.
The Question for Builders
The right question for anyone building AI-enabled enterprise software right now isn’t “what AI features should we add?”
It’s this: what part of our current value is in the scaffolding layer, and what part is in the judgment layer?
The scaffolding layer is whatever your product does that AI will do acceptably well, for free or near-free, within the next eighteen months. Summaries. First drafts. Structured analyses of unstructured inputs. Pattern matching across known frameworks. If your moat is built on being the place buyers come to get those outputs, the moat is already gone. You just haven’t felt it yet.
The judgment layer is where the work requires continuous operation. Domain-specific interpretation. Accountability. Context that only accumulates through sustained engagement with real problems.
That’s where human expertise continues to matter. Consultants, domain specialists, experienced practitioners — their role isn’t going away. The specific outputs they get paid for are shifting.
For vendors, the strategic implication is clear. The scaffolding layer is the cost of admission. Not the product. Get people into your system through that layer as efficiently as possible. Often that means giving it away. Concentrate your actual investment in the judgment layer behind it.
The Question for Buyers
If you’re evaluating AI-powered tools right now, the implication is different.
“How good is the AI output?” is becoming a weaker diagnostic. Most vendor outputs in a given category will be roughly comparable within a year. Asking vendors to differentiate on output quality alone is asking them to differentiate on something that’s already converging.
The better questions are operational.
What happens after the output? How is the underlying data kept current? How does the system behave when a regulation changes at eleven p.m. on a Tuesday and a supplier swaps a material the following Friday? What evidence does the system require before it tells you you’re in a given state? Who’s accountable when something slips through?
Those questions surface real differences. They also surface real gaps in vendors who built around the output and haven’t thought through what continuous operation requires.
When a vendor gives you a valuable AI output for free, the right question isn’t “what’s the catch?”
It’s “what are they confident enough to give away, and what does that tell you about where their defensibility actually lives?”
A vendor who gives you the snapshot for free is telling you the snapshot isn’t the hard part. A vendor who won’t is telling you something too.
The AI shift in enterprise software isn’t about replacement. It’s about redistribution. Value is moving away from the discrete output and toward the system that keeps the output true over time.
For builders, that’s the question worth sitting with.
For buyers, it’s the question worth asking out loud.
