Legal tech is not the market

Two numbers, from the same research house so they can be compared honestly.

Global legal services was worth about $1.16 trillion in 2026 and is forecast to reach $1.38 trillion by 2030. Global legal technology is forecast to reach $69.7 billion in 2033, growing at 12.2% a year.

Put those side by side. The amount the legal services market is expected to grow by between now and 2030 — roughly $217 billion — is more than three times the entire size legal technology will reach in 2033.

Legal tech is not the market. It is a thin layer of tooling sold to the people who serve the market. That framing was correct for thirty years. AI is ending it.

The wall is already coming down

When we catalogued 264 products from 95 legal AI companies, the thing that surprised us was not what the software does. It was how many of these companies are not software companies.

Nineteen of the 95 sell legal work directly. Some were built that way: Crosby reviews contracts as a licensed firm, Keith and Conveyd do English conveyancing, Lawhive and Moritz sell flat-fee legal services. Others started as vendors and added a firm: Norm Ai runs Norm Law, Eudia runs Eudia Counsel, EvenUp runs the pre-litigation department outright.

The economics are not subtle. If AI does most of the labour, licensing software into a $70 billion market is a smaller business than capturing revenue in a $1.2 trillion one. Arizona's alternative business structure licence made it legal for non-lawyers to own the firm doing the capturing, and companies took the opening within months.

We are not describing a trend that might happen. It is in the data now.

The self-represented market barely exists yet

Here is the part we care most about, and where we are least neutral.

Of those 264 products, fourteen are sold to someone handling a legal problem without a lawyer. Thirteen of the fourteen are services, not tools. The one substantial exception we found — Courtroom5 — had been filtered out of our own survey for having raised too little to register.

The obvious read is that the demand is not there. We think the obvious read is wrong, and confuses demand with transactions. Most people facing a legal problem do not buy a cheaper lawyer. They do nothing. The filing goes unmade, the will unwritten, the title unchanged, the hearing attended unprepared. That is not an absent market. It is a market priced out of existence.

One measure of the gap: by our reading of Judicial Council of California data, roughly 71% of California family, probate and unlimited-civil hearings between April 2023 and March 2026 produced no verbatim record at all — three million hearings out of four point two million. Not a worse record. None. Those are proceedings where what was said is simply gone.

Nobody is choosing that. They are living with it because the alternative costs more than they have.

Accessible legal work grows the market

The standard objection is that cheap AI legal services will deflate the market: same work, lower prices, smaller pie.

We think the opposite is more likely, for the same reason cheap photography did not shrink photography. When the cost of doing legal work correctly falls far enough, work that was never going to happen starts happening. The undone filings become filings. And each one that surfaces a real complication produces something the pure-software thesis misses: a person who now needs an actual lawyer, and knows precisely what for.

That is the hybrid we are betting on — self-service at the entry point, a human professional attached at the moment the matter genuinely needs one. Not AI replacing lawyers. AI turning non-consumption into consumption, and handing lawyers the part that deserves them.

Where this could be wrong

Three ways, and they are real.

Regulation. Unauthorised-practice rules still bind most of the country. Arizona and Utah opened; forty-eight states did not. If that stays fixed, the ceiling on all of this is low.

Overreach. The FTC's 2025 order against DoNotPay is the cautionary case. Tools that oversell what they can do for an unrepresented person invite exactly the enforcement that sets the category back.

Deflation might win. If AI compresses billing faster than it unlocks new work, legal services revenue shrinks and the $1.2 trillion prize shrinks with it. We are betting on volume beating price. That is a bet, not a finding.

Why we build what we build

IQdoc starts at the record, because the record is the part of a case that cannot be reconstructed later and the part most likely to be missing. It is also, in our catalogue, one of the least-served categories in the entire market: three products out of 264.

The wall between legal services and legal technology is coming down. Almost everyone walking through it is heading toward the firms and the corporate legal departments, where the budgets already are. We are walking the other way, toward the people who currently get nothing, and we think that side gets larger.

Market figures are third-party forecasts, not our own, and forecasts are often wrong. Nothing here is legal or investment advice.