Small steps, real results: AI in legal case management

By Dale Rounce, Co-Founder of InTouch

I recently hosted a webinar with Tom Lyes, Founder and CEO of Tom Lyes Consultancy Ltd, on how AI is being used in legal case management today. We focused on where AI is already useful for conveyancing and private client teams: data extraction, matter updates, document summaries, drafting support and safer ways to reduce everyday admin.

AI is constantly discussed in legal at the moment, and I understand why some firms feel cautious.

A lot of the conversation is still too broad. It can feel too big, too risky, or too far removed from what fee earners are doing on a busy Tuesday afternoon when they have clients chasing, documents to review, estate agents asking for updates and a task list that keeps growing.

When we talk about AI inside InTouch, we start with the work firms are already doing every day. The useful question is where AI can safely remove some of the repetitive admin that slows teams down, while keeping the person responsible for the matter in control.

Inside InTouch, firms can use AI in two main ways: AI Data Extraction and AI Assist.

AI Data Extraction helps pull key information from documents and move it into the matter. During the webinar, I showed how this can work with a memo of sale, where information such as the estate agent, vendor, purchaser, other side solicitor and property address can be extracted and used to help create the matter.

I also showed how the same approach can be used with a mortgage offer. InTouch can extract the relevant information from the document and use it to help populate a mortgage report, giving the team a faster starting point while still giving them the opportunity to check the information before anything is sent to the client.

This is the type of AI use case I like because it is close to the work. It is not abstract, and it is not trying to replace the fee earner's judgment. It is helping with the admin-heavy part of the process: reading a document, finding the same key pieces of information, entering them into the matter and using them in the next step.

AI Assist works slightly differently. It looks at the information on the matter and helps with drafting, summaries and matter-status questions.

A simple example is an estate agent asking for an update. Instead of the fee earner having to read back through the matter, check the task list and draft the response from scratch, they can ask AI Assist to help prepare a draft based on the information held on that matter.

The fee earner still reviews it, changes it, checks the tone and decides what should be sent. The value is in getting to a decent first draft faster, especially for the updates and messages that take a few minutes each but add up across the week.

The data we shared in the webinar reflects how firms are actually using AI Assist in InTouch. The biggest use is email drafting and tone refinement, which accounts for 34% of use. Document summarisation accounts for 18%, and matter status or “what’s next” questions account for 14%.

These are normal, everyday tasks. Draft an email to the client. Chase an update. Summarise an email chain. Pull out the key points from a document. Check what is outstanding on the matter. Ask whether searches have been received.

None of that sounds dramatic, but it is where time gets lost.

One of the reasons I prefer AI inside the case management system is that it can work from the matter. AI Assist is not being used as a general open tool where someone has to copy information out of the file and paste it somewhere else. It is working with the information already held on the matter, so the team can keep the work closer to the file and review the output in context.

However, control is still essential, and this is where we talked about the 10-80-10 rule.

The first 10% is the person. You define the task, give the context, set the tone and explain what you want. If the input is vague, the output will usually be poor.

The 80% is where AI can help with the heavy lifting. It can draft, summarise, structure and spot patterns. This is where the time saving sits.

The final 10% is the review. The person checks it, refines it, applies judgement and decides whether the output is good enough to use. This is where experience really counts. A good lawyer or experienced conveyancer will know when something feels wrong, when more checking is needed, or when the AI output should be ignored completely.

That is the guardrail I would encourage firms to use, whether they are using AI inside InTouch or anywhere else. Treat the output as a draft. Check it properly. Keep the person responsible for the file in control.

We also spent time in the webinar looking at the wider role of case management, because AI only works well when the rest of the system supports the way the team needs to work.

InTouch has always been built around helping firms reclaim time by automating the robotic tasks, while staying flexible enough for firms to work in their own way. The dashboard is a good example. Rather than starting the day by reacting to the inbox, teams can see which matters need attention, which deadlines are coming up and where new documents have been uploaded.

That gives AI a better place to sit. If the matter information, documents, tasks and deadlines are already organised inside the case management system, AI can support the work around those items more effectively.

For firms looking at AI, I would start with the tasks that already create low-level friction for the team. Manual data entry. Matter updates. Document summaries. Drafting routine emails. Checking what is outstanding. These are the places where small improvements can quickly become useful.

AI in legal case management should feel controlled, useful and close to the matter. Used well, it can help teams move faster without losing the judgment, supervision, and care that legal work still needs.

To see how AI works inside InTouch, book a demo with our team.