Perplexity launched Computer for Counsel on June 24. It is a workflow tool for legal teams, available inside the company's existing Enterprise and Max subscription plans. The product is live now.
The legal AI market has been moving fast. Harvey, one of the best-known purpose-built legal AI companies, raised $200 million in March at a valuation of $11 billion. Legora, a European rival, raised $600 million in April. Anthropic expanded its legal tools in May. Now Perplexity joins that group, though with a different angle.
Perplexity is not building a legal database. It is explicit about that. The product sits as a layer on top of tools lawyers already use: Microsoft Word, DocuSign, Box, SharePoint, and a legal research service called Midpage that provides US case law, statutes, and a tool to check whether a cited case is still valid. The system connects to over 400 applications and can pull context from all of them in a single workflow.
The mechanics work like this. A lawyer assigns a task, say reviewing a third-party NDA. The system breaks that task into steps, pulls the document from the firm's file storage, checks it for risky clauses, fills in entity information, and queues it in DocuSign for signature. Every step links back to a source. The lawyer reviews the output and approves before anything moves. Judgment stays with the human.
This is the core pitch, and it addresses something real. Nearly 75 percent of lawyers say administrative tasks are a significant time drain, according to a Thomson Reuters survey. The work that Computer for Counsel targets, contract intake, citation checks, regulatory monitoring, document drafting, is exactly the kind of work that consumes hours without requiring a law degree to execute.
The pricing angle is significant. Harvey costs an estimated $60,000 or more per year per seat, designed for large firms with big budgets. Perplexity's Max plan is priced far below that, which positions Computer for Counsel squarely at smaller law firms and in-house legal teams at companies, the groups that cannot afford enterprise legal AI but still need something better than a generic chatbot.
The accuracy question matters in legal more than almost anywhere else. Courts fined lawyers $145,000 in the first quarter of 2026 alone for submitting AI-generated briefs that cited cases that did not exist. Perplexity's answer to this is source-linked outputs: every claim connects to its source, and the lawyer verifies before the work product leaves the firm. The Midpage integration handles the specific check of whether a cited case is still valid law, which is a non-negotiable in legal work.
There are real limits. Several connectors, including Clio and Ironclad, are listed as coming soon. The product depends on Midpage for case validity checks, which means the citator coverage is only as good as Midpage's database. And Perplexity is still dealing with unresolved copyright lawsuits over how it sources web content, a background risk that legal teams will want to note.
For in-house legal teams at non-law businesses, the most relevant use case is probably regulatory monitoring and contract triage. A small legal function at a mid-sized company spends significant time tracking whether new laws apply to them across different countries or states. Computer for Counsel can build and maintain that kind of dashboard automatically, with citations attached.
The wider picture is that general-purpose AI companies are entering legal from every direction. OpenAI, Anthropic, Microsoft, Palantir, and now Perplexity have all made deliberate moves into the sector. The legal teams that started piloting tools two years ago are now ahead. Those that have not started are evaluating in a market that is no longer early and is getting more expensive at the enterprise end.