Designing Your ‘Matter Copilot’: AI as a Junior Team Member

The most useful way to think about AI in case management is not as a separate “AI tool”, but as a junior team member attached to each matter — a “matter copilot”.

Like a good trainee or paralegal, your copilot should:

help you keep track of what is going on;
draft simple documents and notes;
spot potential issues and gaps;
never send anything out without supervision.

This article looks at how to design your own matter copilot in practical terms, so that AI feels like a junior team member working inside the file rather than an extra website to juggle.

1. Define what your copilot does — and does not do

Start by deciding what you actually want from a matter copilot. Common, realistic capabilities include:

summarising email threads and documents in context;
suggesting tasks and deadlines from correspondence and orders;
building and updating chronologies;
drafting first-pass internal notes and short client updates;
suggesting time entries from recent activity.

Equally important is what the copilot does not do:

it does not send emails to clients or courts on its own;
it does not give legal advice or decide case strategy;
it does not change the matter record without a human approving the change.

Writing this down — even as a simple one-page internal spec — keeps expectations realistic and aligned.

2. Make the copilot live inside the matter view

For the copilot metaphor to work, AI needs to show up where lawyers already work:

in the matter overview;
alongside the email and document lists;
next to the task and time entry panels.

Good patterns include:

a “Copilot” panel that shows recent suggestions (tasks, notes, time entries) derived from the last few days’ activity;
inline “help me with this” buttons next to emails, documents and notes that trigger specific AI workflows;
outputs that appear as draft items inside the matter (for example, a draft note or task) ready for human editing and approval.

If your copilot lives in a separate tab or website, it will feel like yet another system, not part of the team.

3. Base suggestions on the live matter record

A matter copilot should not rely on whatever individual users remember to copy and paste. Instead, it should draw on:

emails and documents already linked to the matter;
existing tasks, deadlines and time entries;
basic matter data (parties, court, key dates).

This allows the copilot to:

summarise “what has happened since we last updated the client”;
propose new tasks when an order or email contains deadlines;
highlight apparent inconsistencies or gaps (“hearing date mentioned in email but not yet in the diary”).

The more complete your matter record, the more useful the copilot becomes — just like a trainee who keeps a good running note.

4. Design supervision into every interaction

To keep risk under control, every copilot action should have a clear human checkpoint, for example:

suggested tasks appear in a review list where the fee-earner can accept, edit or reject each item;
draft notes and updates are marked as “AI-assisted” until a solicitor approves them;
time entry suggestions are grouped and editable before posting to the ledger.

You can also build simple controls such as:

limiting certain actions (for example, drafting client-facing advice) to more senior users;
requiring explicit sign-off for key outputs (for example, anything that might be disclosed to a court).

The goal is a copilot that behaves like a keen trainee: proactive, but always under supervision.

5. Give the copilot a memory — but only for the matter

A good copilot remembers:

what has already been done;
which prompts work well for this matter or team;
decisions and preferences you have recorded.

Technically, this means:

storing AI outputs and key prompts as part of the matter record;
allowing the copilot to refer back to earlier summaries and chronologies;
avoiding a single, uncontrolled “global memory” that mixes data from different clients.

Matter-level memory helps with consistency (“use the same way of describing the client’s business as before”) while respecting confidentiality and data separation.

6. Start small and expand capabilities over time

You do not need to launch a fully-formed copilot in one go. A sensible sequence might be:

Phase 1 – summaries and chronologies
- email thread summaries;
- document summaries;
- matter chronology suggestions.
Phase 2 – tasks and time
- task extraction from correspondence and orders;
- time entry suggestions from recent work.
Phase 3 – drafting
- internal notes;
- simple client updates;
- cover emails and short letters based on templates.

Each phase can be piloted with a small group, with prompts and workflows refined before wider rollout.

Where OrdoLux fits

OrdoLux is being designed as a matter-first platform where the copilot idea comes to life:

AI features are available directly from the matter view, next to emails, documents, tasks and time entries;
suggestions (summaries, tasks, time entries, notes) appear as drafts inside the matter for human approval;
all copilot activity is logged at matter level, making supervision and audits straightforward;
matter-level memory allows the copilot to build on previous outputs without leaking information between clients.

In other words, OrdoLux aims to give each file a quiet, diligent junior team member — present on every matter, consistent in behaviour, and always under your control.

This article is general information for practitioners — not legal advice or specific implementation guidance for any particular firm.

Looking for legal case management software?

OrdoLux is legal case management software for UK solicitors, designed to make matter management, documents, time recording and AI assistance feel like one joined-up system. Learn more on the OrdoLux website.