AI Is Coming for the Billable Hour — and the Public May Be the Biggest
AI is not merely entering law. It is beginning to change the economics of legal power.
OpenAI’s Astra for Law is the latest signal.
The system combines a frontier model with a dedicated index of U.S. case law, statutes, regulations, court rules and administrative decisions, specialized legal instructions, and plugins connecting the model to professional legal systems. (OpenAI Help Center)
A few years ago, professional AI largely depended on building specialized applications around models that could not reliably perform enough of the underlying work.
That architecture is changing.
Increasingly, the frontier model is capable enough that specialization can live around it:
authoritative data + institutional knowledge + rules + permissions + workflow + governance.
For specialized SaaS companies, that is a significant warning. Model companies are moving upward into their application layers.
For law firms, however, there is a much bigger consequence.
The economic unit of law is under pressure
Roughly 90% of U.S. legal dollars still flow through hourly billing arrangements, according to Thomson Reuters’ 2026 State of the U.S. Legal Market report. (Thomson Reuters)
That creates an obvious structural conflict.
AI’s fundamental productivity proposition is:
Do more work in less time.
The traditional billable-hour proposition is:
Revenue increases with time.
Both cannot remain perfectly aligned.
This is why the disruption is larger than faster legal research.
AI attacks the billable-hour economics that have allowed delay, repetition, opacity and administrative friction to remain profitable.
The ABA’s ethics guidance already recognizes the tension. An attorney billing hourly generally charges clients for the time actually spent using and reviewing AI-assisted work — not the hours the same task might once have required. (American Bar Association)
The scale does not need to be dramatic to matter.
Consider one lawyer billing 1,800 hours annually.
As a simple scenario — not an industry forecast — assume 25–35% of the work can be materially compressed and AI cuts the time associated with those tasks by 50–60%.
That removes roughly 225–378 hours from the traditional billing equation.
Approximately 200–400 hours per lawyer, per year.
Multiply that across a firm and the economics change very quickly.
The productivity does not necessarily disappear. Firms can take additional matters, move toward value pricing, concentrate attorneys on higher-value judgment, or retain some of the efficiency as margin.
But the old assumption that hours themselves represent value becomes increasingly difficult to defend.
The other side of the equation is the client
This is where the story becomes much more important.
Legal systems have always contained an enormous information asymmetry.
The lawyer understands the terminology.
The insurance company understands the process.
The employer has counsel.
The institution knows the system.
The party able to purchase more professional time can purchase more research, more document processing, more analysis and more ability to withstand the process.
The ordinary person frequently enters scared, confused and dependent on someone else simply to explain what is happening.
AI does not make that person a lawyer.
But it can allow them to:
- summarize a large case file;
- identify missing documents;
- translate legal language;
- compare competing arguments;
- prepare questions before meeting counsel;
- and understand whether their matter is actually moving.
That changes the individual’s position.
They become considerably harder to keep powerless through complexity.
Money will still purchase better counsel.
Relationships will still matter.
Sophisticated institutions will have sophisticated AI too.
AI does not create equal legal power.
But it begins to compress one important source of inequality:
the price of understanding.
That matters in a system where the Legal Services Corporation reports that low-income Americans receive inadequate or no legal assistance for 92% of substantial civil legal problems. (Justice Gap)
It also matters because Americans are already going to court without lawyers.
The National Center for State Courts reported in 2025 that the share of state-court cases involving at least one self-represented litigant has risen from roughly 4% in the 1990s to 55%, with substantially higher rates in some case types. (National Center for State Courts)
The opportunity is therefore not simply convincing more people to go pro se.
It is giving millions of people who already cannot afford full representation a better way to navigate the space between:
Google it yourself
and
retain a lawyer for every hour of work.
The legal market begins to unbundle
AI creates the potential for a much larger lawyer-light market.
A person can do more preparation independently.
Then buy professional judgment where professional judgment matters.
That could produce a very different service architecture:
AI-assisted understanding
→ client preparation
→ limited-scope professional review
→ negotiation or advocacy
→ full representation when necessary
This does not necessarily shrink the legal market.
It may expand it.
Millions of people who cannot afford conventional full representation may suddenly be able to afford some professional legal intervention.
But that requires firms to rethink what they sell.
If AI can conduct significant research, review documents, organize records and produce competent first drafts, then charging clients primarily for the production process becomes increasingly difficult.
The human premium moves upward.
Toward:
strategy
judgment
advocacy
negotiation
relationships
accountability
outcomes
The new premium firm
This also changes legal marketing.
The future winner will not differentiate itself by announcing that it “uses AI.”
Everyone will.
The meaningful differentiation becomes operating performance.
Clarity. Velocity. Judgment.
Clarity means the client knows where they stand.
Velocity means the case moves.
Judgment means experienced lawyers concentrate their time where human expertise materially changes the outcome.
Thomson Reuters reports that 71% of in-house professionals already expect law firms to change their commercial models as AI adoption grows, while only 28% of firms report having changed pricing in response. (Thomson Reuters)
That gap will not hold indefinitely.
The legal firms with the strongest future position will not necessarily be the largest or those with the most AI subscriptions.
They will be the firms that redesign the operating model around the new economics.
And underneath that sits the same principle State & Signal applies to every AI transition:
Hedge the technology. Own the intelligence.
Models will change.
Applications will change.
Today’s dominant legal AI system will eventually be replaced by something better.
A firm’s proprietary knowledge, decision logic, client understanding and institutional judgment should survive those shifts.
The firms that understand that will use AI to eliminate low-value friction while strengthening their highest-value human capabilities.
The public should demand the same thing.
Because efficiency alone is not the goal.
The efficiency must eventually reach the person paying the bill.
The future of premium legal service will not be measured by how many hours a firm can bill.
It will be measured by how clearly it can see, how quickly it can move, and how well it can judge.
.