Client/Firm Privilege Boundary Orchestrator for Brisbane Boutique Firms: Keep Matter Files, Firm Knowledge, and AI Tooling on the Right Side of Privilege

You run a seven-lawyer Brisbane firm. Two partners ran a wills dispute last year; now a different client wants advice in a related corporate matter. Someone — a paralegal, an associate, you — drops a draft into the firm’s AI assistant to summarise it. The model has seen the prior matter’s notes because they were ingested into the firm-wide knowledge base. The output references facts you cannot ethically be holding for this client. That’s privilege bleed, and at boutique scale it’s almost structural: you don’t have the walls a top-tier firm builds out of habit, and the AI tooling cuts across whatever walls you do have.

The problem

Boutique firms run AI tooling across small teams where everyone touches everything. The same generative tools that make a seven-lawyer firm competitive against a fifty-lawyer one also collapse the silos that confidentiality and privilege depend on. Three failure modes recur:

Cross-matter leakage — a model trained or retrieved over the firm’s full document store surfaces facts from Matter A while a lawyer drafts on Matter B.
Client-to-firm bleed — client material gets pulled into firm-internal templates, prompt libraries, or “lessons learned” stores that are then used on unrelated clients.
Expert evidence contamination — where the firm briefs an expert (or acts as an expert in tribunal proceedings), AI-assisted drafting risks crossing the independence boundary the Administrative Review Tribunal expects of expert evidence.

The Australian Solicitors’ Conduct Rules — Rule 9 (confidentiality) and Rule 10 (conflicts concerning former clients) — are written for human memory and physical files. They apply identically to AI-mediated information flows, but most boutique firms have no orchestration layer that enforces them at the tooling level.

What the Client/Firm Privilege Boundary Orchestrator does

The orchestrator sits between your lawyers and any AI tool they use. It manages privilege boundaries between client work and firm-internal use of AI by tagging every input and output with the matter, client, and privilege class it belongs to, and refusing retrievals or completions that would cross a boundary the firm has declared.

Concretely, it provides:

A matter-scoped context layer — prompts and retrievals are confined to the matter the lawyer is working in, unless explicit cross-matter authority is recorded
A firm-internal vs client-confidential classification on every document and prompt
Former-client conflict checks before any new matter is allowed to access the shared knowledge layer
An audit trail showing what model saw what content under which matter, exportable for ASCR compliance review or a conflict challenge
Boundary rules for expert-evidence workflows, where independence of analysis must be demonstrable

How it works

Matter intake — every new matter is registered with a client identifier, privilege class, and conflict-check result against the firm’s former-client register
Tagging at ingest — documents, file notes, and draft prompts are tagged at the moment they enter the system, not retrospectively
Retrieval boundary — when a lawyer queries an AI tool, the orchestrator restricts the retrievable context to the active matter plus firm-internal material explicitly classified as boundary-safe
Output check — completions are scanned for references to entities, facts, or phrasings that originate outside the active matter’s permitted scope; flagged outputs require lawyer acknowledgement before use
Audit log — every prompt, retrieval, and completion is logged with matter, user, and boundary decisions, retained for the period the firm’s risk policy specifies

Why this matters in Brisbane

Brisbane’s boutique sector — wills and estates, family, commercial litigation, tribunal-facing practices — works heavily across the Administrative Review Tribunal and Queensland courts. The ART’s practice directions on expert evidence set expectations for expert independence and the basis of opinions tendered to the Tribunal; AI-assisted preparation that silently incorporates material from unrelated matters undermines both the independence expectation and the privilege of the source matter. For a boutique, a single privilege complaint to the Queensland Legal Services Commission consumes weeks of partner time that a firm your size cannot absorb. The orchestrator is built so that the boundary exists in the tooling, not only in the heads of the people using it.

Sources

Administrative Review Tribunal — Practice Directions and Other Guidance: https://www.art.gov.au/help-and-resources/professionals-and-practitioners/practice-directions-and-other-guidance
Law Council of Australia — Australian Solicitors’ Conduct Rules: https://lawcouncil.au/policy-agenda/regulation-of-the-profession-and-ethics/australian-solicitors-conduct-rules
Office of the Australian Information Commissioner — Guidance on privacy and the use of commercially available AI products: https://www.oaic.gov.au/privacy/privacy-guidance-for-organisations-and-government-agencies

Exegesis capability references:

Join the waitlist

Join the waitlist — be the first to know when access opens for Brisbane boutique firms

The Client/Firm Privilege Boundary Orchestrator is in build. We’re scoping the right deployment model for firms under ten lawyers — single-tenant per firm, with the boundary rules expressed in language a managing partner can read and a conflict-check officer can audit. Join the waitlist and the conflicts and boundary patterns you describe to us will shape what ships first.