← The Intake
The Intake · Weekly briefing

This week in legal AI

New York's bar says lawyers should not bill clients for the time AI saves, as an AI-native firm sells fixed-fee deal work and bidders circle a collapsing Big Law firm's playbooks.

Week of October 31 – November 6 2026
Category Market intelligence
Reading time 5 minutes
01 · The week at a glance

AI is repricing routine legal work, and the bar, new firms and vendors are each claiming a share

October 31 · Governance
AI software costs count as overhead unless tied to a matter and disclosed.
November 2 · Structure
No deal values or headcounts have been disclosed.
02 · Governance

New York's bar tells lawyers not to bill clients for the time AI saves them

The New York State Bar Association's governing body adopted a report from its AI committee on Saturday, October 31, as the ABA Journal reported on November 3. It wants limits on what lawyers can bill for time saved by AI, or for time spent learning the tools. The ABA Journal's account does not say what the limits are, and does not mention flat fees.

Limited
Billing for hours AI saved, or hours spent learning the tool
The report calls for limits. The reported summary does not set them out.
Passable
The cost of AI software, as a disbursement
Only when tied to a specific client matter, and with the basis for the charge disclosed. Otherwise it is overhead.
No notice needed
Administrative AI use such as spell checks, calendaring and timekeeping
As long as the use does not affect fees.
Still required
Supervising lawyers stay responsible for review
A firm-approved tool or policy does not remove the need to check each matter's work.

This is the other half of last week's Clio data, where a vendor reported AI-using firms billing more hours. Bar guidance is a nudge, not a price list, and it binds New York lawyers rather than the firms you instruct elsewhere. But it gives in-house teams something concrete to point to when a bill for routine work looks the same before and after AI.

The Flank read

A rule against billing for saved time treats the symptom. The cause is inexpensive work done by expensive resources: an NDA or standard redline still lands on an hourly desk, so every efficiency gain becomes an argument about the invoice. Taking that work off the hourly desk entirely, to supervised agents with a lawyer reviewing finished output, ends the argument rather than policing it.

03 · Market

An AI-native law firm now sells fixed-fee deal work, claiming prices 30-50% below what clients paid before

Athena raised a $2.5M pre-seed round led by a16z Speedrun, announced November 3 (Artificial Lawyer reports $2.4M). Its founders are former Legora employees, and its affiliated firm, Athena Circle, handles commercial real estate purchase agreements, due diligence and loan documents. Athena's agents do the drafting and checking, and its attorneys approve everything before it reaches the client.

Step 1
Agents work the deal
Check agreements against deal terms, flag inconsistencies, prepare redlines, draft client emails.
→
Step 2
Attorneys decide
Experienced attorneys review the work and make the legal calls.
→
Step 3
Client receives
Nothing leaves the firm until an attorney approves it. Fixed fee per transaction.
30-50%
Lower fee than clients paid on recent deals, per the company
>70%
Cut in attorney review time over the first six weeks, reportedly
<48h
Turnaround on one early purchase agreement
22
Clients signed two months after launch
Source: TNW contributed article, November 3, which TNW notes was not produced by its newsroom. All figures are company-reported.

The model is the point: a firm selling the finished outcome at a fixed price, rather than software sold to firms that keep billing hourly. What the announcement does not show is a client's own before-and-after, the deal sizes behind the 30-50%, or what review looks like on a contested clause. Twenty-two clients in two months is an early signal, not a track record.

The structural question for enterprise legal teams

If a new firm can price standard deal documents at a fixed fee, what stops your outside counsel doing the same? Ask your panel which of your routine matters they would quote as a fixed price today.

04 · Structure

Bidders want a collapsing Big Law firm's playbooks and precedents, because AI works from written-down expertise

Weil Gotshal is winding down after partner departures and failed merger talks, according to Above the Law and others, and Artificial Lawyer reports that AI-native firms Vetta and Scissero are now bidding for its talent and assets. Vetta says it wants Weil's precedents, playbooks and knowledge-management infrastructure. Artificial Lawyer adds that Crosby has announced interest.

What the bidders say they want
  • Precedents and playbooks
  • Knowledge-management infrastructure and related technology
  • Lawyers who want AI at the centre of their practice
What has to be checked first
  • Whether a precedent carries any client-specific information
  • That confidentiality is preserved for Weil's former clients

No bid values or headcounts are public, and Vetta's founder, a former Latham & Watkins lawyer, is describing his own bid. Still, the target is telling. The scarce asset in an AI-run practice is the firm's written standard for how routine work should be done, not the model reading it. Artificial Lawyer notes that a precedent stripped of client detail may be reusable, but that any buyer has to confirm it.

The Flank read

Your own playbooks, templates and escalation rules are the same asset, and you already own them. Big Law spent decades pricing the application of standard precedents at partner and associate rates, which is inexpensive work done by expensive resources. Supervised agents that work from your documents make that routine work cheap, and you keep the knowledge.

05 · Data

Microsoft says lawyers accept over 70% of Copilot for Legal's suggested edits, but not how that was measured

In an Artificial Lawyer interview published November 3, Microsoft's legal vertical lead described Copilot for Legal, which launched October 15 and appears automatically for people signed up to Copilot. Microsoft says suggested review changes see acceptance rates above 70%, and that edits are retained at a rate in the high seventies. It plans further public benchmarking, but has published no results.

Suggested changes accepted70%+, per Microsoft
Edits retained"high seventies", per Microsoft
Sample size, task mix, playbooks usedNot stated
Price and licence tierNot stated
Source: Artificial Lawyer interview, November 3. Figures are Microsoft's and interviewee-reported. Bar lengths show the stated percentages only.

An acceptance rate measures how often a reviewing lawyer clicks accept, which depends on who is reviewing and what they were reviewing. It is a useful usage signal, not an accuracy rate. The design also assumes a lawyer reads every suggestion inside Word, so the work stays with the person doing the reviewing. Microsoft says it tests whether reviews against the same playbook give consistent results, which is the right question, but no results are out.

The structural question for enterprise legal teams

For your standard agreements, what share need no lawyer edit at all? A vendor that can only report how often edits are accepted is telling you that a lawyer still reads everything.

06 · So what

Routine legal work is being repriced, and the open question is who keeps the saving

A bar association wants limits on billing for AI-saved time, a new firm sells the same work at a fixed fee, bidders want a failed firm's playbooks, and a platform vendor reports edit acceptance. Everyone is moving to capture the value of routine work. Few are saying who reviews it.

Put AI savings in your outside counsel guidelines
New York's report gives you a reference point for asking what happens to time AI saves.
Ask your panel for fixed-price quotes on routine matters
AI-native firms are now quoting them, so the comparison exists.
Ask vendors for no-edit rates, not acceptance rates
What share of your standard agreements needed no lawyer touch at all?
Treat your playbooks as an asset
Written, current and owned by you: it is what bidders are chasing at Weil.
The Flank view

The week's stories share a gap. Routine work is still priced as if it were complex, so AI savings become a dispute over the bill rather than a lower cost. Inexpensive work is done by expensive resources until the routing changes.

Outsource legal work to supervised agents. Flank's agents know your templates, terms and escalation rules, take the NDAs, standard redlines and procurement agreements, and hand finished work to a named lawyer for review before anything leaves.

Subscribe

The Intake

Weekly briefings on what's actually changing in legal AI: the market shifts, regulatory moves, and structural questions that matter for enterprise legal teams. Written by the Flank team.

Subscribe on Substack
Flank

Outsource legal work to supervised agents

Enterprise legal teams use Flank to handle high-volume contracting end-to-end: NDAs, MSA redlines, procurement, triage. Agents that know your templates, terms, and escalation rules. Lawyers review finished work.

Learn more at flank.ai