← All essays

E/04 · The guide

AI agents for lawyers: the working layer of an AI-native practice

This is what AI-native looks like in practice: not a chatbot you remember to ask, but agents configured to use your matter data as context, running first on every matter while you review. Each agent handles one job in the workflow, mapped to the MATTER Method's six primitives.

By Raghav R Handa · Founder · July 2026 · 11 min

Why agents, not just AI

A general chatbot does not know your client's name. A matter-wired agent does.

A general AI tool needs the full context pasted in every time: client name, matter details, jurisdiction, deadlines, documents. That is not automation, it is extra work. A matter-wired agent is pre-configured to query the practice's own databases. The intake agent knows what a new matter looks like in your system. The research agent knows the jurisdiction of the matter you are working on. The time agent knows what you logged today. Context is built in, so you get the output, not the setup work.

The agents

Every agent. Every job it does.

Matter · Onboarding

Intake agent

Turns a new client inquiry, a message, an email or an intake form, into a fully structured matter record: client details, matter type, jurisdiction and initial action items.

SavesAbout 45 minutes per new matter

Research · Analysis

Research agent

Takes a question, the jurisdiction and the relevant facts from the matter and returns a cited research summary you can verify. A first-draft memo in minutes.

Saves2 to 4 hours per research task

Documents · Correspondence

Drafting agent

Drafts letters, notices, agreements and standard documents using the matter's own context, client name, dates, parties and applicable law, for your review.

Saves1 to 2 hours per document

Time · Billing

Time recording agent

Reviews the day's activity across your matter records and drafts time entries in your preferred billing format. No reconstructing the day from memory at 6pm.

Saves30 to 60 minutes daily

Evidence · Review

Evidence agent

Reads documents uploaded to the matter, contracts, affidavits, exhibits, and extracts key facts, flags inconsistencies and highlights provisions that need attention.

Saves1 to 3 hours per document review

Client · Communication

Client update agent

Reads the current state of a matter, open actions, recent developments, upcoming deadlines, and drafts a status email in your voice.

Saves20 to 30 minutes per update

Actions · Planning

Action generator

Looks at the matter type, jurisdiction and current stage, then generates the next set of actions needed. A structured checklist built from how the work actually unfolds.

SavesAbout 30 minutes per matter review

Invoicing · Revenue

Billing agent

Reviews logged time for a matter, cross-references the fee agreement, and drafts an invoice narrative with itemised descriptions ready for approval.

Saves1 to 2 hours per invoice cycle

Risk · Review

Compliance checker

Runs a document against a configurable checklist, conflicts, confidentiality clauses, governing law, and flags issues before it leaves the office.

SavesAbout 45 minutes per contract review

Correspondence · Routine

Correspondence agent

Drafts routine outgoing correspondence, confirmations, acknowledgements, follow-ups, from templates linked to the matter.

Saves15 to 20 minutes per letter

Knowledge base · Capture

Knowledge agent

When a matter closes, extracts the key lessons, precedents and strategies into the practice's knowledge base, so the next similar matter starts smarter.

SavesTurns every matter into an asset

Every one of these agents ships inside MatterOS and LexOS, and each of them is on the menu in an adnah transformation, deployed only against actions that have already been named, inside explicit roles, with the evidence rule enforced. No unnecessary subscription layers, no per-seat tax on your own system.

Questions, answered plainly

What are AI agents for lawyers?

Pre-configured AI assistants wired into a legal workflow. Unlike asking a general chatbot a question, an agent uses the actual matter data, client name, case type, deadlines, documents, as context, so it can act on a specific case rather than answer in the abstract.

How do AI agents fit into an AI-native practice?

An AI-native practice runs AI first, on every matter, systematically, with the lawyer reviewing and taking responsibility. Agents are how that becomes routine rather than aspirational: each agent maps to one primitive of the MATTER Method, so the machine does the first pass on every matter and the lawyer supervises.

How are matter-wired agents different from a general chatbot?

A general chatbot has no knowledge of a specific practice. A matter-wired agent is configured to use the practice's own data, matters, clients, actions, documents and time entries, as context. An intake agent knows to pull the client's name, matter type and jurisdiction from the record. A general chatbot only knows what gets pasted in manually, every time.

Do I need technical skills to use AI agents in legal practice?

No, when the agents are pre-built and pre-connected. You activate them within the workspace, no code, no API keys, no configuration, and most lawyers have agents running the same afternoon. In an adnah transformation they arrive installed, against your own named actions.

Are AI agents in legal practice ethical and compliant with bar rules?

Yes, when used correctly. Agents assist, they do not decide. Every output is a draft for lawyer review, and the lawyer remains responsible for the final work product. The agents are designed to augment judgement, not replace it.

Which agent saves the most time for solo lawyers?

Time recording and client updates save the most time for most solo lawyers. Manually logging time entries and drafting weekly status emails are the two tasks that eat the most non-billable hours in a small practice. Most lawyers report saving one to two hours a day once these run automatically.