Towards agentic ubiquity
February’s tech and legal tech news demonstrated repeatedly that GenAI, and especially Anthropic, is really good at dominating global media headlines. Some of the stories were hilarious, like Anthropic’s response to OpenAI’s plans to introduce ads in ChatGPT Free and Go subscription tiers. While OpenAI published assurances that ads won’t influence ChatGPT responses, Anthropic’s Super Bowl ads (created by Mother London, if you haven’t seen them, definitely worth a look) asks, well what if they did? The straplines are provocative, and all the ads sample the line from Dr Dre’s 1999 What’s the difference “What’s the difference between me and you?” to make the point that Claude is ad-free.
And some were momentous. February saw the biggest tech news for over a year when the release of Anthropic’s Cowork plugins, which include a legal plugin for contract review, NDA triage, compliance workflows and legal briefings, provoked a massive reaction from global markets.
Plugin fall-out
Tech and legal techs crumbled in what the financial press called the SaaSpocalypse; the markets’ knee-jerk reaction was to sell off tech stocks due to the threat of agentic AI in the form of the Cowork plugins disrupting the software as a service (SaaS) business model: if companies could use agentic AI to build their own tools and automate processes, they would need fewer if any subscriptions to external services. Hence the impact on DocuSign, Salesforce, Shopify and Adobe, among others. This was extrapolated to legal tech stock too. As has been widely reported, Thomson Reuters stocks plunged 16%, LexisNexis’ parent company RELX lost 14%, Wolters Kluwer fell 13% and LegalZoom dropped nearly 20%. The rout deepened on 6 February when Anthropic launched Claude Opus 4.6 with autonomous multi-agent coordination capabilities.
But as often happens in the tech sector, things weren’t what they seemed initially, and this turned out to be a perceived threat, at least for now. It became clear that Anthropic was not looking to push out SaaS companies, or compete directly with incumbent legal tech and legal AI suppliers, but rather to connect to them – I am classifying Harvey and Legora as incumbents because while they haven’t been around for long, they are dominating the market in legal agentic AI. On 24 February, Anthropic announced at its Enterprise Agents briefing that Claude Cowork would integrate directly with SaaS apps. And the briefing included Thomson Reuters CEO Steve Hasker and a surprise video featuring a preview of the next generation of TR’s CoCounsel legal agent which will use Claude’s Agent SDK as its infrastructure. This was swiftly followed by LexisNexis announcing Anthropic legal plugin integrations into its Protégé GenAI platform. This and subsequent supplier integration announcements helped tech and legal tech stocks begin to to recover.
However, this is a significant development, because the plugin integrations are at engineering level rather than at application level – as Chad Atlas put it in his review of the Cowork legal plugin, the difference is that “Anthropic [is] building infrastructure not portals – MCP for productivity, skills for packaging expertise, memory for maintaining context.” The legal plugin depends on MCP (model context protocol), i.e. it’s a two-way connection between the systems, not an API (application programme interface). So effectively it becomes part of the system. Given the number of new integrations, it is clear that Anthropic is looking for ubiquity for its agents rather than market dominance. And ubiquity is expensive, so everybody is fundraising.
All bets on agentic AI
Even before the markets started to recover – and they have not fully recovered – venture funds rushed in to fund agentic AI. It all happened very suddenly, but looking back over the last couple of weeks, it is more logical than it seemed at the time. And legal is a microcosm of the bigger picture.
February saw the GenAI giants raising even more funding. On 12 February, Anthropic raised $30bn in a Series G funding round, valuing the company at $380bn. And on 27 February OpenAI closed $110bn at a $730bn pre-money valuation, more than double the company’s $40bn financing last year.
Legal AI funding is also accelerating. Agentic legal platforms Legora, which is built on Claude, and Harvey, which is built on OpenAI, but also uses Claude as a foundation model, made it clear that they saw the legal plugin as an opportunity not a threat. And it prompted additional investment rounds. Legora is finalising a Series D funding round to raise $400m at a $5bn-$6bn valuation and Harvey is reportedly in talks to raise $200m at an $11bn valuation in a Series G funding round led by Sequoia Capital and GIC. Other lawtech funding focused on AI-native law firms. London-based Lawhive, which was the first lawtech firm to acquire a regulated law firm, raised $60m in series B funding to expand its US business and in the US Y Combinator invested in AI-native firms, including slack-first start-up General Legal which may have been the first law firm featured on Product Hunt.
The UK government too is betting on legal AI. Last week, Lord Chancellor David Lammy announced an additional £4.5m in funding for LawtechUK (£1.5m a year for the next three years) which has been delivered by Legal Geek and Codebase since 2023, as well as significant investments in courts technology and digitisation. According to LawtechUK’s latest investment snapshot, investor confidence in UK lawtech remains high, hitting a record £189m in 2025, with some 47 UK-founded companies winning funding. The number of UK lawtechs increased 17% year on year, and 12 were acquired. So after all the various headlines, even governments are betting on agentic AI.
War of words – or Skynet?
In the US government involvement in GenAI has very different implications, and what started as a war of words escalated as the global context switched to real conflict. When Anthropic signed a $200m contract with the US Defense Department last year, it sought guarantees that the Pentagon would not use its systems for autonomous weapons or mass surveillance of US citizens. But on 24 February, defense secretary Pete Hegseth demanded unrestricted access and when Anthropic refused, President Trump posted on social media that he was directing all US federal agencies to cease use of Anthropic’s technology. And on Friday night OpenAI CEO Sam Altman posted on X that OpenAI had reached an agreement to work with the US Department of War. He too referred to safety principles around “domestic mass surveillance and human responsibility for the use of force, including for autonomous weapons systems” forming part of the agreement. But will those principles be adhered to or will we end up with Skynet?
Legal Geek is hosting four conferences this year, learn more on our events page.
Written by Joanna Goodman, tech journalist
Photo credit (Joanna): Sam Mardon
