Agents of change

Does the 'agentic' AI law firm spell the death knell for the billable hour? This new legal-practice model is reshaping not just the tools that lawyers use, but the very structure of how legal services are being delivered. Blathnaid Martin eyes up the future

The legal profession has long been characterised by its alleged conservatism, its attachment to precedent, and its wariness of transformative change.

Yet a new model of legal practice is emerging, one that is reshaping not just the tools that lawyers use, but the very structure and economics of how legal services are delivered.

This is the rise of the agentic law firm, in which AI doesn't just assist lawyers, but essentially operates as a semi-autonomous agent in legal workflows, capable of executing complex, multi-step tasks with minimal human oversight.

For some more seasoned practitioners, the age of AI may just feel like the latest trend they've experienced throughout their careers, alongside Tipp-Ex, Dictaphones, and word-processors.

Sometime in the 1990s, law firms became digitised. Most practitioners gained access to computers, and firms migrated to using document-management systems, shelving physical libraries and printed books in favour of electronic databases – eventually abandoning fax machines for the convenience of Microsoft Outlook.

The vanishing

The 2010s brought the era of process automation, and law firms started using contract-lifecycle management software and e-discovery tools. All ultimately improved efficiency, but didn't necessarily revolutionise legal practices.

These technologies all performed discrete, rule-bound, specific tasks, which were human operated.

Instead of letters, lawyers sent emails. Instead of physically thumbing through hundreds of folders of discovery, lawyers searched, scrolled, and clicked through online databases.

Given this history of technological advancement and the slow adoption by the legal sector of new tools and practices – not to mention the fact that lawyers, by their nature, are sceptics – it can be hard to decipher whether AI will actually change legal practice or whether it is simply propaganda, spun up by those who are financially self-invested in AI's success.

Many lawyers questioning how significantly AI could threaten their job security will have already dabbled in using generative AI in practice, whether for research or basic drafting – usually plugging questions into an AI-powered chatbot.

The results can be both impressive and underwhelming and, more often than not, disappointingly 'AI sounding'.

This, of course, offers comfort and assurance to those fearing replacement – a confirmation that AI isn't necessarily faster and definitely isn't capable of doing your job for you – right?

Eyes without a face

Most dubious lawyers may not have yet used agentic AI. 'Generative AI' is inherently passive: it will do what you tell it to, but much of its results will be dependent on how clear your instructions are, what context you provide, and what information it has access to.

You will need to check its sources. It might offer you suggestions about what it should do next, but you will have to manage it on a task-by-task basis. Prompting generative AI is a skill in and of itself. It can be time consuming and can take a while to master.

Agentic AI, however, can operate completely autonomously.

It can complete complex, multi-step tasks with limited human supervision. Agentic AI can read your emails, download draft contracts and review comments, prepare a redline, and draft a response.

It can run conflict checks, schedule meetings, take attendance notes, and raise invoices. It can continuously monitor for legislative and regulatory updates across multiple jurisdictions and alert you to new changes.

It can do so much more, and it can likely do it faster, better, and cheaper than many lawyers can. Terrifying.

While the age of AI brings insurmountable anxiety to some, the general consensus seems to be that the role of the lawyer won't become entirely redundant and the profession will, instead, adapt with the times.

It may seem less savoury, perhaps, than the introduction of email, but there are benefits that come with the superpowers of agentic AI.

It is likely that the comparative advantage for lawyers will lie in tasks that require creativity, problem-solving, ethical reasoning, and the exercise of judgement.

Much of what lawyers do requires intrinsic human cognition and strategic thinking, of which AI is not (yet) capable.

Crafting a novel legal argument for a court case, weighing up whether or not to agree to a liability cap, or navigating the political dimensions of a regulatory investigation – these are scenarios in which the human lawyer remains indispensable.

Agentic AI also raises questions of professional responsibility and ethics. Solicitors and barristers are subject to duties of competence, confidentiality, and candour that do not permit delegation without sufficient oversight.

Regulators, including the Solicitors Regulation Authority in England and Wales, are beginning to grapple with the implications of agentic AI for existing codes of conduct, but comprehensive guidance remains in its early stages.

Recently, Ms Justice Eileen Roberts of the High Court advised that the Irish judiciary was developing a detailed practice note to guide lawyers on the use of AI in litigation.

It is clear that when an AI agent drafts a letter of advice or reviews a contract, the supervising lawyer remains professionally responsible for the accuracy and completeness of that work.

Regulatory bodies and the judiciary are likely to seek to limit the use of AI in a way that doesn't degrade the profession or put its future economic viability at risk.

At present, at least, it's clear that human lawyers will remain a requirement for the purposes of accountability and responsibility for maintaining sufficient supervision in the age of AI.

Undercover agent

Lawyers in private practice may also find AI slightly more ominous than their in-house counterparts.

This varies depending on the industry, but most in-house legal teams in Ireland will already have been using agentic AI for some time and will have access to specific legal tools, with the likes of Harvey, CoCounsel (by Thomson Reuters), and Legora dominating the legal ops market.

This isn't surprising, with the tech sector itself being characteristically primed to embrace and adopt these technological advancements.

There is also more pressure on in-house legal teams to be lean, agile, efficient and to scale with a rapidly growing business.

The AI crunch is leaving in-house legal teams with less budget to hire more lawyers, while companies will have more appetite to invest in AI services that promise cost savings over time.

There is a rising pressure to be more efficient, and agentic AI may seem like a gravy train you either have to board or best get off at the platform.

This likely differs from the experiences of many law firms, where legal services have historically been monetised by the billable hour model, which ties revenue directly to the number of hours worked.

This model creates well-documented perverse incentives: rewarding inefficiency, discouraging investment in technology, and making it difficult for firms to scale without increasing headcount.

The inherently conservative character of law firms and their low appetite for risk, married with a usually tight budget for legal ops, and this sort of revenue model that capitalises on inefficiencies might provide great comfort to a nervous practitioner contemplating AI.

It may well be the case that lawyers in private practice don't embrace AI at the same speed that in-house counsel do, but this is unexpectedly causing a gap in the market and giving rise to new firms who are taking advantage.

Invasion of the body snatchers?

Crosby AI, which launched its business in January 2025, describes itself as an "agentic law firm built for execution". Crosby utilises AI agents to speedily review contracts that may ordinarily take law firms weeks to return comments on, boasting a median review time of just 58 minutes.

They use qualified lawyers (including professionals from Harvard, Stanford, and Columbia) to oversee the review and charge their clients on a fixed-fee basis for each contract, no matter how many rounds of comments the contract goes through.

The firm advertises that it "combines the speed and intelligence of AI with the safety of lawyers-in-the-loop to review contracts in under an hour". The idea is to align the firm's financial incentives with those of its clients: closing deals quickly.

Charging significantly less and moving significantly faster, they are already starting to disrupt the legal market for commercial contracts.

Some will point to the fact that many clients would not have the risk appetite for this approach and will want to stick with traditional big-name firms who boast experience, longevity, and 'top-tier' status.

A quick scroll through Crosby's client list, however, names most of the who's who in emerging AI. They don't need to eat the whole pie to be disruptive – they can still change the market, bite by bite.

The invisible man

Another example is Radiant Law, a similar style of law firm that offers contract review on a fixed-fee basis. It has been around much longer (2011), but has recently begun utilising agentic AI.

It's a popular choice in the tech sector for companies grappling with large-scale contractual amendments due to legal and regulatory changes, for example, changes required due to the recent Digital Operational Resilience Act (DORA) in the EU.

This type of business model is, of course, geared towards using AI for contract review, which most law firms are already tinkering with.

Many will point to the fact that this is an area primed for agentic AI adoption: standardised, routine, repeatable, and based on precedents developed from market-wide approaches to certain terms.

Many will also point to the fact that other more advisory or analytical areas are not so vulnerable to competition from emerging AI businesses.

There are emerging players though – for example, Vanta – that describe themselves as a "leading agentic trust platform", offering businesses compliance checklists, draft policies, and continuous monitoring for some of the more process-heavy legal frameworks, such as GDPR, the EU AI Act and HIPAA.

While Vanta's customers may still need initial legal advice to assess their obligations under each of these frameworks, it is clear that new AI businesses are coming up with creative ways to enter the market.

While the impact on legal services may not be immediate, law firms will need to consider how agentic AI will change their revenue models. AI is already being utilised for some of the more tedious tasks that trainee solicitors or junior lawyers traditionally manage, such as large scale e-discovery or due-diligence projects.

As more senior lawyers begin to adopt agentic AI and firms start to compete with new businesses offering fixed-fee work, many may need to assess if charging on the basis of billable hours incurred is fair, reasonable, and maintains a competitive advantage.

Could agentic AI be the death of the billable hour? Perhaps for some areas of legal services, firms will reassess how they capitalise on work and how AI could prove helpful in reducing costs. But like almost everything 'AI' right now, only time will tell.

Blathnaid Martin is a practising solicitor and in-house legal counsel at MongoDB, a publicly traded US tech company with over 70,000 customers, 7,000 employees, and over $2 billion in annual revenue. She holds a master's in EU law from King's College London.