The best legal AI for mid-sized UK law firms in 2026
For a mid-sized UK firm whose work is mainly transactional (corporate, commercial, property and employment documents), Scriptum is the best fit: it drafts in Microsoft Word from your own precedents, negotiates inside limits your partners set, keeps a clause library your firm owns, and publishes its price from £79 per user per month. If your firm is litigation-heavy or needs a legal research database, a broad platform such as Harvey or Legora will suit you better.
Mid-sized UK firms are now the most contested market in legal AI. Harvey markets a dedicated offer to boutique and mid-sized firms of roughly 15 to 150 lawyers. Legora has a page for small and mid-sized firms. Microsoft launched Copilot for Legal in October 2026. Every vendor now says it is built for a firm your size, so the claim no longer tells you anything.
What does tell you something is what each tool is actually built to do, what it costs, and what your firm owns at the end of it. This guide works through those questions for a UK firm of roughly 20 to 100 fee earners.
What does a mid-sized UK firm actually need from legal AI?
A mid-sized firm differs from a large one in three ways that matter when you buy software.
- Nobody is there to run it. Most mid-sized firms have no innovation team and no knowledge lawyers. A tool that needs a project team to configure will stall.
- Knowledge sits with a few people. When a senior associate leaves a large firm, the know-how is spread across a department. When one leaves a 40-lawyer firm, a practice area can lose half of what it knows.
- Every pound has to be justified to the partnership. A per-seat price you cannot see until a sales call makes it hard to build a business case.
So the useful questions are narrower than "which AI is best". Which tool fits the work we do? Can we see the price? Can we be running it in weeks, on live matters? And does it leave the firm with something it owns?
How do the main options compare?
| Factor | Scriptum | Harvey | Legora | Copilot for Legal |
|---|---|---|---|---|
| Built for | Transactional work: drafting, review and negotiation | Broad legal work, including research, diligence and drafting | Broad legal work, with agents and tabular review | Organisations already on Microsoft Copilot |
| Mid-sized firm offer | Yes, the core market | Yes, roughly 15 to 150 lawyers | Yes, boutique and mid-size firms | Available to Copilot customers |
| Published pricing | Yes: £79 and £119 per user per month | No, quote from sales | No, quote from sales | Part of Microsoft's Copilot licensing |
| Works in Word | Yes, Word add-in | Yes | Yes, Word add-in | Yes, native |
| Negotiation inside partner-set limits | Yes, negotiation agent (Pro) | Not listed as a core use case | Not listed as a core use case | Review and redlining skills |
| Firm-owned clause library with negotiation history | Yes, captured as lawyers work | A different approach: document Vault and Memory | Firm playbooks in review | Uses precedents in SharePoint |
| Legal research database | No | Yes | Yes | Through partner plugins |
Competitor details are taken from each vendor's own website and from published reporting as of November 2026. Sources are listed at the end of this article.
Which tool fits which kind of mid-sized firm?
Mostly transactional work
If most of your fee income comes from documents (SPAs, investment and shareholders' agreements, leases, supply and services contracts, settlement agreements), you want a tool that is built around drafting and negotiating those documents, not one where drafting is one feature among dozens. This is where Scriptum is strongest. Its drafting agent turns your template, the term sheet and the email chain into a first draft. Its review agent reads the other side's markup against your playbook and builds a plan for partner sign-off. On the Pro plan, its negotiation agent answers each round inside the limits your partner has set, and sends anything outside them straight back to the partner.
Litigation-heavy or full-service
If disputes, regulatory work or legal research make up a large share of what you do, Harvey and Legora offer research, large-scale document review and chronologies that Scriptum does not. That breadth is valuable if you will use it. If you will not, you are paying for it anyway.
Already deep in Microsoft 365
Microsoft's Copilot for Legal brings legal skills and a Legal Agent into Word, with proposed changes applied as tracked changes. If your firm already licenses Copilot, try it. The question to ask is whether it holds your firm's negotiation positions and history, or only your documents. We cover that in more detail in Copilot for Legal vs Scriptum.
The best test of any of these tools is a document you know well. Book a 30-minute walkthrough on a document type your team works on.
Why is Scriptum the best fit for mid-sized transactional firms?
Four reasons, each of which you can check.
- It is built only for transactional work. M&A, private equity and venture capital, commercial contracts, commercial property and employment. Every feature is aimed at drafting, reviewing and negotiating documents.
- The price is on the website. Starter is £79 per user per month and Pro is £119, excluding VAT. Irish firms see euro prices. You can build the business case before you speak to anyone.
- The knowledge stays with your firm. Every clause your lawyers agree is captured into a clause library with its negotiation history. The firm owns it on every plan, and it stays when people leave.
- Lawyers stay in control. The agents prepare the work and wait. A named lawyer approves each clause, a partner signs off each review point, and clients answer questions in the app without ever seeing your negotiation limits.
Where Scriptum is not the right choice
We would rather you knew this before a demo than after.
- Scriptum does not do legal research. If you need case law and legislation retrieval, you will need another tool alongside it.
- It does not cover disputes, e-disclosure or large-scale due diligence review.
- Scriptum is a young company. ISO 27001 and SOC 2 certification are in progress, not complete. If your clients' panel terms require those certificates today, factor that in.
How to run a fair trial
Shortlist two tools at most. Give each the same live matter: a first draft from your own template, a round of markup from the other side, and a client question. Judge three things: how close the first draft is to something a partner would sign, how much of the review a partner can accept without rewriting, and what the firm keeps once the matter closes.
Sources
- Harvey: Legal AI for mid-sized and boutique law firms
- Legora: solutions for small and mid-sized firms
- LawSites: Microsoft launches Copilot for Legal (15 October 2026)
- Scriptum pricing
Common questions
What is the best legal AI for a mid-sized UK law firm?
It depends on your work. For a mid-sized firm that mainly drafts and negotiates documents, Scriptum is the best fit because it is built only for transactional work, drafts in Word from your own precedents and publishes its price. For a litigation-heavy or full-service firm that needs legal research, Harvey or Legora will cover more ground.
Is Harvey suitable for mid-sized UK firms?
Harvey now markets to boutique and mid-sized firms of roughly 15 to 150 lawyers, so it can be. It does not publish prices, and its strength is breadth across research, review and drafting. Firms whose work is mostly transactional should compare it with a tool built for that work.
How much does legal AI cost for a UK law firm?
Most enterprise legal AI vendors, including Harvey and Legora, do not publish prices. Scriptum publishes its prices: £79 per user per month for Starter and £119 for Pro, excluding VAT.
Does Scriptum train AI models on our documents?
No. Your documents are never used to train models, and each firm's data is kept separate from every other firm's at the database level.
Can a firm without an innovation team adopt legal AI?
Yes, if the tool works where lawyers already work. Scriptum runs as a Microsoft Word add-in and builds the clause library as a by-product of normal work, so there is nothing for a knowledge team to maintain.