Where a firm loses hours
A lawyer’s time is sold by the hour or is part of a fixed fee, and in both cases the firm cannot sell hours spent on routine to a client. Typical losses:
1. First inquiries: listen, see whether the case fits your practice, explain which documents are needed.
2. Archive search: where the firm had a similar dispute, which position it took, which template to use.
3. The first draft of a contract, a claim letter or a lawsuit, which a senior lawyer edits anyway.
4. Reading the case file and building a timeline before a hearing.
5. Tracking procedural deadlines across dozens of cases.
A separate loss: inquiries that come in the evening and at weekends, while lawyers are in court or offline. A client who gets no answer goes to another firm, and the hour of meeting prep is no longer needed.
In its Future of Professionals report of June 2025, Thomson Reuters gives a savings benchmark: 5 hours a week per professional. Lawyers already look for these hours on their own: according to Clio Legal Trends 2025, 79% of lawyers use AI, and 87% in large firms.
Five processes for AI
1. Client intake: AI answers the client in a messenger or on the website, collects the gist of the question, documents and deadlines and books a consultation. It gives no legal opinion.
2. Search for case law and the firm’s positions: an answer from the archive with links to the documents it comes from.
3. Drafts of contracts, claim letters and lawsuits on the firm’s templates, with facts from the case file.
4. Review of incoming contracts: deviations from the template and risky terms, with clause numbers.
5. Deadline tracking: a summary across cases and reminders to the responsible lawyer.
Each process has its own AI agent with its own rights: the intake agent cannot see the case archive, and the case law search agent does not write to clients.
Large firms took the same path, from simple questions to processes. At Allen & Overy, lawyers asked AI about 40,000 questions in testing, and only then was the service opened to 3,500 lawyers.
A firm with a few lawyers should start with one of the five processes, the one with the most repetition and a result that is easy to check.
How to count the savings
We count savings in hours, not in impressions. The calculation needs four numbers:
1. The number of lawyers who do the task.
2. The hours a week one lawyer spends on the task now.
3. The share of the work AI takes on: drafting, search, checking.
4. The cost of a lawyer’s hour to the firm, or the client rate.
The product of the first three numbers gives the hours freed per week, and multiplying by the hourly cost gives the money. You can run your own numbers in the lawyer time calculator.
A sample case: 4 lawyers spend 6 hours a week each reviewing incoming contracts. If AI takes on half of this work, the firm gains 12 hours a week, about one and a half working days of one lawyer.
The Thomson Reuters benchmark of 5 hours a week per professional is useful for checking the result: if your calculation gives much more, review the assumptions.
The firm sells the freed hours to clients or spends them on complex cases. In the pilot, we check this number on your cases and write the metric into the contract.
Security and confidentiality
The firm’s AI runs on the firm’s server or in a cloud in Russia. Personal data is processed under 152-FZ, Russia’s personal data law, and document text does not go abroad or to services without a data processing agreement.
AI has a separate account and rights only to its own process: the intake agent does not need access to the case archive. All actions are logged.
AI does not follow commands from incoming messages. Decisions, signatures and negotiations stay with the lawyers.
The firm needs an AI use policy: which services lawyers may use, what they must not upload and who checks the result. According to Clio Legal Trends 2025, 53% of lawyers say their firm has no such policy or they do not know about one.
Rollout plan
1. A free 30-minute review: we choose the process and the metric.
2. Pilot: one AI agent on one process from $1,499 in 14 days, with the success metric in the contract.
3. AI department: 3–5 AI agents on several processes from $4,999 in 4–6 weeks, with CRM, accounting system and telephony connections and staff training.
4. Support from $480 a month: monitoring, edits and a metrics report.
The price and term are fixed in the contract before the start, and the pilot cost counts toward the price of the department.
After the pilot, the firm gets a report on the metric from the contract and decides for itself: stop, keep the agent on support or add processes. The agent with a perpetual license, the data and the access stay with the firm in any case.
If the firm has several practices and needs to see where AI gives the most hours, start with the AI Solutions Map: a process analysis with an effect calculation from $93,000 in 7 weeks.