For years, law firms have adapted the way they work to fit the software they buy. Anthropic’s latest release suggests that relationship is beginning to change, with potentially significant consequences for the business of law.
Buying legal software has traditionally involved accepting somebody else’s idea of how a law firm should operate. A practice-management system arrives with its own workflows, fields and assumptions. The firm trains its people, adjusts its procedures and develops workarounds for whatever the software cannot accommodate. Over time, those compromises become part of the way the practice operates.
Anthropic’s latest development suggests that this relationship may be starting to reverse.
On 28 October, the company introduced a new collection of applications within Claude for Legal that allow lawyers to create and customise tools around their own working practices. Known as Artifacts, they include contract review systems, matter-management dashboards, citation checkers and regulatory monitoring tools. Lawyers can adapt them using ordinary language, connect them to existing information sources and share them with colleagues.
The technology is interesting in its own right, but the greater significance lies in who gets to design the systems through which legal work is performed.
“Legal Tech” has, until now, meant getting the profession accustomed to purchasing technology from specialist suppliers. The prospect of building useful applications around a firm’s own expertise represents a substantial shift. It also opens up an intriguing possibility for smaller firms: access to tailored technology without the development costs that have traditionally put it beyond their reach.
From software users to software designers
The legal technology industry has spent decades developing increasingly sophisticated products for the profession. Practice management, document automation, contract review and legal research have all become established software markets, supported by specialist vendors and significant investment.
Much of this technology works extremely well. However, we have increasingly seen issues arise in the space between a commercial product's standard functionality and the particular way a firm wants to conduct its business.
Every practice has its own methods, developed through years of experience. A probate solicitor knows the points at which an estate administration is most likely to stall. A conveyancing team has established procedures for managing outstanding requisitions. A commercial lawyer knows which contractual provisions regularly generate negotiations and which departures from the firm's preferred position require partner approval. That knowledge is valuable, but translating it into software has generally required technical expertise and a budget sufficient to justify the development. Anthropic is making that translation considerably more accessible.
Take its new Playbook Deviation Board. The application compares contracts against an organisation’s preferred terms, identifies departures and directs exceptions to the appropriate person for approval. A legal team can adapt the tool to reflect its own negotiating positions and approval procedures. The underlying task is familiar. What has changed is the ease with which practitioners can begin shaping the application to reflect the way they actually work.
A solicitor can describe the required process in plain English, supply the relevant precedents and identify the exceptions and refine the resulting application through further instructions. The lawyer's expertise becomes part of the design of the system rather than something that must be accommodated within a predetermined set of functions. This has obvious implications for efficiency, but it may also change how firms think about the value of their accumulated knowledge. At Praxis we have always advocated that starting with the work is the way to win at legal tech. Now, with this release, procedures developed over many years can easily become repeatable, measurable processes available to the wider practice. And because the cost of experimenting with these applications is falling, firms can begin addressing operational problems that would never have justified a conventional software development project.
An opportunity for smaller Irish practices
For smaller Irish law firms, this could prove particularly valuable. Consider the work involved in taking on a new client. Initial enquiries must be assessed, relevant information collected, conflicts checked, documents requested and matters assigned. The precise sequence varies according to the practice area and the firm's procedures. Much of the work is administrative, yet mistakes or omissions at this stage can create difficulties throughout the life of a matter.
A firm that understands this process could design a tailored intake application to collect the appropriate information, identify missing details and direct matters for professional review. The system could reflect the firm's own approach to particular categories of work, rather than requiring the practice to adopt a generic process.
Similar opportunities arise in probate administration, conveyancing, litigation support and commercial contract review. These are areas in which relatively small operational improvements, repeated across hundreds of matters, can generate worthwhile savings.
There is also a potential competitive advantage for smaller firms. Large practices have traditionally enjoyed greater access to bespoke technology, internal innovation teams and substantial implementation budgets. As the technical barriers to developing applications fall, the ability to identify useful improvements and implement them quickly becomes more important.
A smaller practice may have an advantage here. Its partners often have a close understanding of how work moves through the office, where delays arise and which tasks consume disproportionate amounts of professional time. Decisions can be made quickly, and new processes tested without the organisational complexity of a large firm.
The opportunity is to use that knowledge more effectively.
There is, however, an important discipline involved. Building an application requires a firm to explain its processes with sufficient precision for a system to follow them. Procedures that have worked informally for years may contain inconsistencies, undocumented exceptions or differences between individual practitioners.
The exercise of designing a workflow can therefore be valuable before any software is produced. It requires a firm to establish how work should be handled, where decisions belong and what standards must be maintained.
The legal technology market is changing with it
Anthropic's announcement also offers an indication of how the wider legal technology market may develop.
Claude for Legal now supports more than 60 legal connectors, including Bloomberg Law, Legora and Filevine. Its new gallery includes applications developed with established law firms and specialist providers. Harvey, for example, has contributed an intake application that allows business users to submit legal requests through Claude, with routine enquiries addressed using approved guidance and other matters directed to lawyers.
The involvement of these companies is significant. Specialist legal technology providers are increasingly integrating their capabilities into broader AI platforms, giving customers new ways to access information and organise work across different systems.
For law firms, this could create greater flexibility in how technology is assembled and deployed. A practice might retain its established practice-management software and research services while building additional applications to address particular workflows or connect information held across several systems.
It also raises questions for the software industry. If a firm can develop its own interfaces and processes, the value of a commercial product may increasingly depend on the quality of its underlying data, specialist functionality, integrations and reliability.
Established providers retain considerable advantages in all these areas. A configurable application is a long way from a fully supported production system, and law firms will continue to need dependable technology infrastructure. But the economics of software development are changing, and both suppliers and their customers will have to adjust.
Managing partners may find themselves making a different calculation when considering new technology. Rather than immediately searching the market for a product that performs a particular task, they may first examine whether the capability can be developed around systems and information the firm already possesses. Overall this represents a meaningful change in the purchasing relationship.
Greater freedom requires stronger governance
The ability to create and customise applications also brings professional responsibilities into sharper focus. A firm developing an AI-assisted contract review process needs to understand how the system identifies relevant provisions, what sources it relies upon and how exceptions are treated. An application handling client intake must be designed with appropriate controls over confidential and personal information. A regulatory monitoring tool needs reliable sources and a process for checking that its information remains current.
These are questions of professional supervision as much as technology management.
The Irish regulatory environment is already moving in this direction. The Law Society's guidance on generative AI addresses the duties of competence, confidentiality and professional responsibility. High Court Practice Direction HC 142, effective since 1 September 2026, requires independent verification of AI-generated material used in court documents.
Anthropic's new citation checking application provides a useful example. The tool is designed to identify inaccurate quotations, incorrect references and legal propositions unsupported by their cited sources. Such functionality could make the review process more efficient, but the practitioner remains responsible for independently verifying the material. Paragraph 13 of HC 142 makes clear that asking an AI system to confirm accuracy is insufficient, subject to the permitted verification of original source material accessed through the system.
As firms begin building their own tools, governance will increasingly need to address the applications themselves: their intended purpose, access to data, testing, supervision, maintenance and responsibility for their outputs.
An application used across multiple matters can reproduce an error just as efficiently as it performs a useful task. Firms must be able to explain how their systems work, where professional judgment is exercised and what happens when the technology encounters a situation outside its intended scope.
These controls are particularly important when applications begin to influence the routine handling of client work. The professional standards expected of a solicitor remain unchanged, regardless of who developed the software.
A different kind of competitive advantage
Richard Susskind has long argued that the future of legal services will be shaped by the systematisation of professional knowledge and the redesign of legal processes. For many firms, these ideas have remained associated with major technology investments, specialist consultants or innovation programmes at larger practices.
The new generation of AI development tools brings the possibility much closer to everyday legal practice.
It also suggests that technological competence will come to mean something more than proficiency with the latest products. Solicitors who can analyse a workflow, define the necessary decisions, identify exceptions and understand where human judgment must intervene will be well placed to influence how technology is developed and used within their firms. This ability to bring together legal expertise, operational understanding and the ability to improve how services are delivered can be a game changer for smaller firms.
Anthropic's new applications are an early stage in this development. Their reliability, limitations and practical value will need to be established through experience. Questions of security, integration, maintenance and dependence on individual platforms remain important, and established software engineering disciplines will continue to have a role.
Nevertheless, the direction of travel is becoming clearer. Law firms are gaining greater influence over the design of their own technology, and the cost of turning professional knowledge into useful applications is falling.
For managing partners, this creates an opportunity to look again at how work is organised, where professional time is spent and which processes would benefit from a more tailored approach.
The next generation of legal technology may be shaped as much by the experience of practising solicitors as by the companies that supply their software.
After years of adapting their practices to technology, law firms are acquiring the ability to make technology adapt to them.
Sources and further reading
- Anthropic | 28 October 2026 — Introducing Artifacts for Legal Work
- Courts Service | 1 September 2026 — High Court Practice Direction HC 142
- Law Society of Ireland — Generative AI Guidance