AI that moves matters: A guide to AI for law firms in 2026
From hype to habit
The hype is real, and it is not going anywhere. AI is now part of everyday legal work, sometimes useful, sometimes muddled and messy. Most firms are not chasing robot lawyers; they are keeping trusted workflows and asking a simpler question: what can AI take off our plate today, safely and without changing who we are?
Yes, there are people paid to plan for the future, testing what “futuristic” AI might do in two to five years, but they are the exception. Small firms in particular do not have the budget for dedicated AI strategists. The vast majority are focused on one thing: what helps us move matters faster and safely today?
The real work now is deciding which tasks to hand over, which to supervise, and which to keep firmly in human hands.
Adoption is no longer theoretical. Across the profession, usage has jumped. The ABA’s latest reporting shows generative AI adoption nearly tripled in a year, from 11% in 2023 to 30% in 2024, with 46% of firms with 100 or more lawyers already using AI tools. In parallel, legal-sector polling indicates a shift from pilots to live workflows. By early 2025, 26% of legal organisations reported actively using generative AI, up from 14% in 2024. For a UK lens, one survey found the share of lawyers using generative AI for work rose from 11% in July 2023 to 41% by September 2024.
These numbers place us at the start of a transition. The technology is impressive, but the old ways will not change overnight. Adoption is here, yes, but getting ahead of ourselves would be a mistake. The tools will improve and firms will adapt, yet the only place to start is the present: use what works on live matters today, prove value, and build from there. Smaller firms are eager to adopt technology, although barriers persist, which is why clear, incremental wins matter.
Meanwhile, the bottlenecks remain stubbornly human: re-keying data from contracts, forms and statements, searching for facts across emails and bundles, and turning first drafts into client-ready prose. Clients expect speed and transparency; manual prep slows both. Whether it is a share purchase agreement, a probate application, a settlement note or a property exchange, the pressure is the same: clear facts, clear drafts, clean data.
The only AI that counts is AI that moves matters. That means three tests, every time: Speed (start from context, not a blank page), Accuracy (pull from source documents and map to fields you can trust), and Control (human review, approvals, and a clear audit trail).
What follows looks at the jobs AI can do now inside the file, where the opportunities sit for fee-earners, senior lawyers, partners, finance and operations, why responsible use is non-negotiable in legal work, and a brief look at anonymised InTouch usage — the prompts teams reach for most right now, and how those habits are helping.
Responsible AI: Guardrails, not guesswork
This is non-negotiable. It is legal work, with clients, money and risk on the line. Responsibility has to sit at the centre. That means using AI where it genuinely helps and slowing down where judgment is needed, not trying to automate everything. Most firms know this, but we are in a transition and it is worth stating clearly.
Human in the loop
Drafts and field mappings are suggestions. You review, edit and approve. Nothing is sent and nothing updates in the file until you confirm.
Auditability
Every action needs a trail in the matter: who asked, what was generated or changed, when it happened and which sources were used. Permissions should control who sees what.
Data handling
Keep processing inside your environment and inside the matter. Use the documents and emails already in the file, minimise copy out to external tools, and apply redaction or access limits when needed.
Risk posture
Sensitive or disputed issues get extra human specificity and sign‑off. For complaints, negligence risk, settlement terms or anything high stakes, start with a first version, add the facts yourself and escalate where appropriate.
If you are assessing case management software, ask three simple questions: Where do drafts and mappings live? Can we block auto‑send? Where is the full log? InTouch has these guardrails built in, so work stays in the matter with your approvals and a complete history.
What can AI do for your firm? Tasks, not features
If you are still deciding whether to bring AI into the firm, you are not alone. Many platforms show the same headline capabilities, so the real question is what will help your team this week on real files without rebuilding how you work.
AI should feel like your current process, only faster and more accurate. In practice, that means three uses you can rely on today: Find the facts, Frame the first draft, and Fill the fields from the documents you already have. These three appear on almost every file, use the material already in the case, and cut time and mistakes without changing your process. You still review the result before anything leaves the file.
Of course, AI can do other things. It can try to draft long documents from scratch, predict timelines from past matters, search case law at scale, build knowledge hubs, route emails, transcribe meetings, or power dashboards. Those may be useful later, but they often need extra data, integrations or policy work, and many of them run outside the matter, which adds copy and paste and slows adoption.
Start with Find, Frame, Fill inside the matter, with review and a clear record of what changed. When that is working well, you can add the rest with a clear head.
Find
Finding should feel like asking an experienced colleague who knows the file, only quicker. You ask a clear question and get an answer you can verify, linked to the source. The system looks across tasks, emails, notes, documents and key dates inside the matter so you move from question to action quickly.
Where AI can help:
Pull answers from the live file with links to the exact page, email or note.
Handle typical asks like “What is outstanding on 14 High Street?”, “Do we have the signed agreement?”, “When did we last contact the other side?”, “Where is the case or policy number?”.
Flag when information is missing and suggest a next step, for example drafting a request.
Log every query and answer; no changes are made until you decide.
Frame
Framing gives you with a draft that fits the matter and the reader. Names, dates and references come from the file, so you edit for judgment and tone rather than assembling facts. This speeds everyday messages and makes harder notes easier to shape without losing control.
Where AI can help:
Produce drafts for client updates, dated chasers, plain English clause explanations and 10 point summaries.
Pull the right facts from the matter so content is specific, not generic.
Let you choose format and length, for example email, letter or SMS.
Keep you in control with review and approval before anything is sent.
Fill
Filling turns information in documents into clean, trusted fields in the matter. Parties, amounts, dates and references are suggested for the right places, with the source shown for quick checking. You approve updates so data stays accurate and consistent across files.
Where AI can help:
Map key data from agreements, court or regulator forms, policies and, for property, Memoranda of Sale and mortgage offers.
Show the origin of each suggestion and hold changes as pending until approved.
InTouch can scan PDFs and phone photos to pull text and fields, with a link back to the right page.
Improve reporting and downstream steps by keeping fields correct, reducing re keying and aligning task triggers and statements.
What AI should not do (and where it is weak)
AI should not replace judgment, send anything without review, update fields without approval, bypass your process or permissions, or move data outside the matter. Those are firm lines.
There are also areas where AI is simply not strong and should be used with care:
Drafting long, complex documents from scratch without full context, templates, and a clear brief.
Predicting outcomes or timelines with certainty, or giving legal advice beyond summarising what is in front of it.
Handling sensitive complaints or disputes where tone, nuance and subtext matter; start from a first version, then add specifics.
Getting numbers exactly right every time; totals, references and figures need a human check.
Citing law or policy without precise sources and dates; always verify the reference.
Reading poor-quality scans or photos; OCR can miss details when the source is weak.
Pulling reliable facts from scattered systems; it works best when everything is inside the matter.
Keeping firm style or precedent on its own; give it examples and review the result.
Who can AI help, and how?
Now comes the question every team asks in practice: what does this mean for the different roles that keep a firm running each day? The answers are not new workflows or big promises but small, steady improvements that keep work moving.
Everyone can see where a matter stands, drafts are ready when you need them, and the data you rely on holds up when you report to partners. Start where the pain shows, keep the review step in place, and let the time you save roll back into client work.
Clients and Cases (fee earners)
AI helps you keep everyone up to date without losing time to searching or retyping. It turns the live status of a matter into clear updates for clients, the other side and third-party intermediaries, adds the next step with a sensible date, and surfaces important items like approval or policy expiries, disclosure timelines and, in property work, search timelines before they slip. Because names, references and dates come straight from the file, you can focus on judgment and tone while the system handles the assembly.
How AI helps most:
Updates go out sooner, so files move without long pauses.
Deadlines and expiries are flagged early, so you act before problems grow.
Handovers are smoother because every message and source sits in the matter.
Knowledge and Drafting (senior and quality)
AI makes long documents easier to work with, turning contracts, policies, witness statements and leases into short briefs that highlight risks, responsibilities and deadlines. It can explain clauses in easy-to-understand English when a client needs clarity, and produce simple checklists or tables that set out what information is needed for standard forms and pre-completion or pre-hearing checks. Partners see the main issues sooner, so reviews move without hold ups.
How AI helps most:
Reading time drops and key points are clear, so decisions come faster.
Risks and dates are surfaced consistently, which reduces rework later.
Shared checklists raise the floor for juniors and keep outputs consistent.
Operations and Management (partners, ops, finance)
AI helps keep matter data consistent and up to date, which makes oversight simpler and planning more reliable. Key fields such as parties, prices or amounts, dates and references are captured in a standard way across files. It also flags billing and accounting mismatches early and produces simple snapshots that show opened matters, ageing files and who is waiting on whom. Compliance is easier to evidence because changes, approvals and sources are recorded in the file history where you expect them.
How AI helps most:
Reports match reality because core fields are complete and consistent.
Billing issues are caught before they reach clients, which protects cash flow.
You can spot bottlenecks quickly, so work is allocated where it will move.
How are teams using AI today? Data from inside InTouch
Matter AI Assist is the AI helper built into InTouch. It runs inside the case workspace, uses your permissions, and saves drafts and field suggestions back to the file for review and approval.
The snapshot below is an anonymised, de-duplicated sample of real prompts used on live files up to 7 October 2025.
These patterns appear across conveyancing, disputes, corporate, employment and private client work.
What we see in practice
Most use is everyday work: check the live status, turn answers into short updates, summarise documents, run simple checks and tidy small workflow steps. Prompts are plain English, outputs stay in the matter, and nothing is sent or updated until someone approves it.
A small sample of the prompts teams are using today…
Status and next steps:
“What is the status of this matter?”
“What enquiries are outstanding?”
“What do I need to do next?”
Drafting and comms:
“Draft an email to the client giving an update.”
“Email to other party suggesting completion date.”
“Turn this into a customer-friendly message I can send via SMS.”
Documents and explanations:
“Summarise the lease.”
“Please provide a summary of costs, fees, or data in this report.”
“Explain limited title guarantee.”
Money and process checks:
“Does the invoice and statement add up correctly?”
“Prepare a completion statement.”
“Please produce a table with relevant info to complete LP-1F form.”
Workflow and reporting:
“Link two matters.”
“How many matters have been opened in the last two months and list types.”
“Reword this paragraph for simpler reading.”
What this means today
Teams use AI in InTouch to move live matters forward, not as a side experiment. Because it all happens inside the file with permissions, audit history and approvals, people save time and keep control.