Voice Preservation & Procedural Fairness Agent for Brisbane Boutique Firms: Keep the Client’s Voice (and the Disclosure Record) Intact

You run a six-lawyer firm in Brisbane. Two of your matters are at the Administrative Review Tribunal — a protection visa review and a NDIS decision. The client’s statement matters: the Tribunal weighs credibility on the words the applicant uses, the sequence they tell their story in, the things they emphasise. Your junior associate has been using an LLM to tidy submissions and rewrite witness statements into “cleaner” English. The drafts read well. They also no longer sound like your client, and there is no audit trail of what the model changed or whether AI use was disclosed under the ART’s practice directions. The Voice Preservation & Procedural Fairness Agent is built for this exact gap.

Why it matters now

The Administrative Review Tribunal publishes practice directions and guidance that govern how matters are conducted before it, including expectations around expert evidence and the conduct of representatives. Where AI is used in the preparation of submissions or evidence, disclosure and accuracy obligations sit alongside the practitioner’s existing duty of candour under Rule 19 of the Australian Solicitors’ Conduct Rules. Two distinct risks land on boutique firms harder than on large practices: first, there is no internal compliance team auditing what tools your associates used to draft what document; second, AI-rewritten client statements can quietly substitute the model’s register for the client’s own — a procedural fairness problem the Tribunal cannot fix on review if the original voice is gone. For ART matters where credibility findings are decisive, a “tidied” statement is not a neutral edit.

The 5-minute view

AI-Use Disclosure Non-Compliance is the failure to record, disclose, or substantiate the use of generative AI tools in preparing material filed with a tribunal or court
The ART’s practice directions and guidance set expectations for representatives appearing before it, including in relation to expert evidence and the conduct of proceedings
ASCR Rule 19 (candour to the court/tribunal) extends to AI-assisted drafting — a non-disclosed model rewrite of a client statement is a candour problem, not a stylistic one
“Voice preservation” means keeping the client’s lexicon, sentence rhythm, and order of recall intact in any AI-assisted edit — and recording exactly what changed
Procedural fairness in administrative review depends on the decision-maker hearing the applicant, not a model’s paraphrase of the applicant
The Voice Preservation & Procedural Fairness Agent compares the original client-authored text against any AI-touched version, flags substantive register shifts, and produces a disclosure-ready change log
The agent runs locally — no client statements are transmitted to external LLMs

What Exegesis is building

The Voice Preservation & Procedural Fairness Agent is a tier-three service shape in the Exegesis Legal stack, built alongside RuleCheck (the open-source pre-lodgement checker at github.com/andrefabre/rulecheck). The agent ingests two artefacts — the original client-authored statement or instructions, and the draft submission or witness statement that AI has touched — and produces a structured diff that distinguishes (a) typographical and grammatical corrections, (b) structural reordering, and (c) substantive voice or content changes. Substantive changes are surfaced for the supervising solicitor to accept, reject, or annotate before lodgement. Every accepted change is logged with a timestamp and the tool used, giving the firm a disclosure record it can rely on if the Tribunal or a regulator later asks how a document was prepared. The agent is local-first and deterministic in its diff layer; it does not generate replacement text.

The deliverable

A pre-lodgement Voice & Disclosure Report comparing the client-authored source against the AI-touched draft
Three-tier classification per change: typographical, structural, or substantive
Substantive changes surfaced for solicitor sign-off before lodgement
A disclosure-ready change log suitable for inclusion in the matter file or, where appropriate, in submissions to the Tribunal
Markdown export for archiving alongside the client’s original statement
Optional integration with RuleCheck for combined citation + voice preservation pre-lodgement checks

CTA

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

The Voice Preservation & Procedural Fairness Agent is in build. We’re scoping pricing (per-matter, per-user, or firm-licence) based on what we hear from boutique firms running ART and merits-review workloads. Join the waitlist and what you tell us will shape the tier you sit in.

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
Federal Court of Australia — Use of Generative Artificial Intelligence Practice Note (GPN-AI) (comparative reference for disclosure expectations in federal proceedings): https://www.fedcourt.gov.au/law-and-practice/practice-documents/practice-notes/gpn-ai

Exegesis capability references: