The pricing on AI title tools is structured to obscure what you actually pay at scale. Here’s the math, by tier, with the gotchas vendors won’t write down — including ours.

The pricing sheets you’ll see from AI title vendors — Autopilot included — generally look like one of three structures. Per-order pricing, where you pay a flat fee per file processed. Per-document or per-search pricing, where you pay for each underlying object. And tiered subscription pricing, where you pay a monthly fee that includes some amount of usage, and overage above that. Every vendor presents whichever of these three makes them look the cheapest at the file count you mentioned in the discovery call.

Here’s how the math actually shakes out.

Per-Order Pricing

The headline number on per-order pricing is usually somewhere between $4 and $12 per file. Looks great if you’re doing 200 files a month — $800 to $2,400 sounds bearable. The gotcha is that “per file” is almost never defined the way you’d define it. Most per-order pricing counts a file every time the system touches it — so a refi that requires a re-search after a recording delay is two files. A file that bounces back from QA is two files. A file that the AI flagged for human review is sometimes billed as two files (the AI pass and the human pass). Read the definition of “order” carefully and ask, in writing, what counts and what doesn’t. If the vendor won’t put it in writing, you have your answer.

Per-Document or Per-Search Pricing

Headline numbers look smaller — sometimes $0.40 to $2 per search or document. The gotcha is volume math. A typical file requires somewhere between 6 and 20 underlying searches and documents depending on the property and the state. Multiply that out and per-document pricing is often two to three times more expensive than per-order on the same volume. Vendors who lead with per-document pricing know this. The other gotcha on this model is that the vendor decides what counts as a billable search — and the incentives are all in the wrong direction. We’ve seen pricing sheets where pulling a deed counts as one search and pulling the same deed a second time (because the first pull had bad OCR) counts as a second search.

Tiered Subscription Pricing

The headline is a flat monthly number with an included volume — say, $2,500 a month for 250 files. Reads clean. The gotcha is the overage rate, which is almost always 1.5 to 3x the implied per-file rate inside the tier. So if you do 250 files, you pay $10/file effective. If you do 300, the next 50 cost you — at typical overage rates — $15 to $30 a file, so your effective per-file rate jumps from $10 to about $12.50. Vendors set tier boundaries deliberately just below typical file counts so the customer is structurally always in overage.

Two Gotchas Across All Three Models

One: minimum monthly commits. A monthly minimum locks you into paying for volume you don’t have during slow months. The bigger problem is that monthly minimums are usually written as annual commits with monthly billing, so even if you’re losing money on them, you can’t exit until renewal. Two: premium-tier feature locks. The AI features that matter most — the actual examiner-assist work — are often gated behind the highest tier, while the basic tier gives you not-very-useful automation that you didn’t really want. Vendors price this way because the upgrade conversation is more lucrative than the new-customer conversation.

Now for Our Pricing

Scribe is priced per seat. That’s per examiner that Scribe assists. Plus actual AI usage fees. There are no monthly minimums and no annual lock; month-to-month is the default. We publish all of this on the pricing page.

The risk of being this transparent, frankly, is that competitors can read this and reprice. Some of them will. The reason we’re publishing the math anyway is that I’d rather have independents in this industry able to evaluate AI title vendors honestly than have us win deals because nobody can compare apples to apples. The market only gets better when the pricing gets readable.

See Autopilot’s published pricing tiers — view the page →