Anthropic has released a legal focused plugin for Claude Cowork and Code, aimed at in house legal teams and workflow style work such as contract review, NDA triage, compliance workflows, litigation briefings, and templated responses. (This is described as automation support, not legal advice, and Anthropic explicitly says outputs should be reviewed by licensed lawyers before relying on them.) (The Guardian)

The immediate impact was not just product chatter. Shares in listed names fell sharply on the news, including Pearson, RELX, Sage, and London Stock Exchange Group. (The Guardian)

For practice, the more useful lens is practical rather than speculative. Anthropic’s open source plugin materials are structured around repeatable “playbook” positions and consistent review steps, including escalation triggers and fallback positions. That aligns with what many firms are rediscovering: the value is less about a single model and more about institutionalising your house approach, in a way juniors can follow and partners can supervise. It also highlights a small firm constraint.

These tools are not a magic button and require technical experise (although modest). If “clone the repo” sounds like a foreign language to you, a DIY plugin approach is still a stretch without support and guardrails and I expect this will be the case for law firms. We are not the most tech proficient profession! (Artificial Lawyer and GitHub)

Takeaways

If you fancy trying Anthropic's new offering, you will require a Claude subscription. I personally used it with Claude Code, rather than Copilot as I do not have Copilot access on my Windows laptop.

Using Claude Code, I used git to clone the repo and asked Claude to set up a working folder on my computer. From there I have been testing the skills and templates (and added a few of my own).

So far I am impressed as a solution for easily repeating prompts that I have tested and saved. It saves me copying and pasting from my Obsidian vault and keeps everything nicely organised.

See below for my current folder structure and a snapshot of that in my CLAUDE.md file, and how I am envisaging structuring this. So far this is not implemented with actual real cases but rather I am using it with dummy data to see if my idea will work and whether I should explore Enterprise licences for Claude.

If this sounds too technical for you, this offering from Anthropic is likely not for you. I promise it isn't hard to implement but you need to be comfortable with command line tools (if you use it with Claude Code). Those of you with access to Claude Copilot will likely fare better but I have been unable to test!

I do think this is a sign of things to come from Anthropic but I think the market overreacted somewhat.

Whilst I do not have access to any of the big legal tech tools at the moment, such as Harvey etc, it seems a key advantage is the "out of the box, all in one" solution these offer. There is no need for often tech adverse lawyers to learn how to use git to clone the repo on Anthopic plugins or use any scary looking command lines.

Always adhere to the Golden Rules!

Read: Artificial Lawyer

In order to help cover the running costs of this newsletter, please check out the advert below. In line with my promises from the start, adverts will always be declared and actual products that I have tried, with some brief thoughts from me.

This week’s advert echos my sentiments related to consistent prompting and AI use. Training is key!

AI implementation often goes sideways due to unclear goals and a lack of a clear framework. This AI Training Checklist from You.com pinpoints common pitfalls and guides you to build a capable, confident team that can make the most out of your AI investment.

What you'll get:

Get the checklist.

Build versus buy for legal AI just got louder (Bloomberg Law)

A useful “legal ops” view on why model providers shipping legal features threatens thin wrapper products, but does not eliminate the hidden cost of implementing and maintaining DIY tools. Helpful reading if you are deciding whether to standardise on a platform or assemble your own workflow stack.

Read or listen: Bloomberg Law

What the legal plugin actually contains (Anthropic, on GitHub)

If the lead story interests you, and you want something tangible to review, the repo shows how playbooks, escalation triggers, and structured redline guidance can be represented, which is useful even if you never deploy the plugin itself. A good artefact for thinking about “house style” and repeatability.

Read or listen: GitHub

A sober take on who is really threatened (Artificial Lawyer)

As always, an excellent piece from AL; a grounded argument that the biggest pressure is on commoditised tools rather than the data heavy platforms, and that adoption still depends on real world integration and governance. Useful if the market reaction has made you jumpy.

Read or listen: Artificial Lawyer

Instead of a practice prompt this week, I would encourage those who have the means and capability to check out Anthropic’s playbook (from the lead story). They have plenty of prompts to try for many professions.

If you do not have access to Claude, consider setting up a similar playbook style folder structure where you can store your approved prompts.

My structure idea (heavily influenced by Claude) is in the lead story.

Consider then making a policy for your business, setting out what prompts are approved for use and what controls will be implemented.

Hit reply and tell me what you would like covered in future issues or any feedback. We read every email!

For future content, I am planning a series of posts about how my firm is assessing options and implementing AI, with true examples and openness about the hurdles we face.

Thanks for reading,

Serhan, UK Legal AI Brief

Guidance and news only. Not legal advice. Always use AI tools safely.

Below are a few newsletters that I recommend, for various reasons. Check them out!