Industries we serve

Built for firms where confidentiality is the work.

We deploy private AI where data sensitivity, regulatory pressure, and professional obligations rule out public AI tools.

Primary focus

Law firms

Drafting, summarising, and research are exactly what AI is best at. Client files and privileged communications are exactly what cannot be pasted into ChatGPT. DeepShield exists to resolve that tension.

Common use cases

·Drafting initial advice letters and engagement letters
·Summarising lengthy judgments and discovery bundles
·Extracting key dates, parties, and obligations from contracts
·Researching precedents within the firm's own matter history
·First-pass document review and redaction suggestions
·Generating plain-English client correspondence

Regulatory fit

ASCR Rule 9.1

Maintaining client confidentiality. Public AI tools transmit data overseas, which regulators have said cannot be done safely. On-premise eliminates the transmission entirely.

Privacy Act 1988 (APP 8)

Cross-border disclosure of personal information requires positive steps to ensure appropriate handling. Keeping AI inference inside Australia removes the cross-border element.

Legal Professional Privilege

Material processed through a third-party AI may lose the confidentiality required to sustain privilege. On-premise inference keeps the chain intact.

The regulator's position

“Lawyers cannot safely enter confidential, sensitive or privileged client information into public AI chatbots or copilots (like ChatGPT), or any other public tools.”

Joint statement: Law Society of NSW, VLSB+C, and the Legal Practice Board of WA. December 2024.

Also serving

Accounting and advisory firms

Client financial records are among the most sensitive data a professional services firm handles. AI can dramatically speed up review, reconciliation, and report writing. But that is only possible if the client data stays inside the firm and never gets uploaded to a server in another country.

Common use cases

·Drafting client-facing reports and management letters
·Summarising financial statements and board papers
·Reviewing contracts and identifying key financial obligations
·Preparing plain-English explanations of complex tax positions
·Generating first drafts of audit findings and recommendations
·Answering staff questions against the firm's own precedent library

Regulatory fit

APES 110 Confidentiality

The Code of Ethics for Professional Accountants requires members to refrain from disclosing confidential information acquired through professional relationships. Uploading client data to a public AI tool is a disclosure.

Privacy Act 1988 (APP 8)

Client financial records routinely contain personal information. Sending that information to an overseas AI provider triggers cross-border disclosure obligations most firms are not meeting.

Client contractual obligations

Engagement letters and service agreements commonly include confidentiality clauses. Using public AI tools with client data may constitute a breach of those terms.

Also serving

Medical and allied health practices

Health information is the most heavily regulated category of personal data in Australia. Clinicians increasingly want AI assistance for documentation, referrals, and patient correspondence. Doing that safely means the AI cannot be hosted by a third party.

Common use cases

·Drafting referral letters and specialist correspondence
·Summarising patient histories and clinical notes
·Generating discharge summaries and care plan documentation
·Transcribing and structuring consultation notes
·Answering clinical administration queries against internal protocols
·Producing plain-English explanations for patient communications

Regulatory fit

Privacy Act 1988 (Health Information)

Health information is a sensitive category under the Privacy Act requiring heightened protection. Inputting it into a public AI tool constitutes a disclosure to a third party and likely an overseas transfer.

My Health Records Act 2012

Healthcare providers have specific obligations around how health information is accessed, used, and disclosed. Third-party AI processing falls outside the permitted uses.

AHPRA professional standards

Registered health practitioners have professional obligations to protect patient privacy. Regulators have not endorsed the use of public AI tools with identifiable patient information.

Also serving

Engineering and technical consultancies

Engineering firms routinely work under confidentiality obligations that explicitly prohibit disclosure of project data to third parties. AI is genuinely useful for technical documentation, report writing, and specification review. The condition is that the data stays inside the firm.

Common use cases

·Drafting technical reports, specifications, and tender responses
·Reviewing project documentation for key obligations and risks
·Summarising lengthy environmental impact assessments and planning documents
·Generating first drafts of design rationale and engineering justifications
·Answering staff queries against the firm's own project knowledge base
·Producing client-facing summaries of complex technical findings

Regulatory fit

Client confidentiality clauses

Most engineering engagement contracts include clauses prohibiting disclosure of project data to third parties. Public AI tools are third parties. On-premise AI keeps the data inside the firm's boundary.

Privacy Act 1988

Project data often includes personal information about landowners, residents, or stakeholders. Handling that data through a public AI tool creates disclosure and cross-border transfer risks.

IP and commercial sensitivity

Designs, methodologies, and technical approaches may constitute trade secrets or commercially sensitive information. Uploading them to a public model creates risks that are difficult to quantify or remediate.

Also serving

Financial planning and wealth management

Financial planners hold some of the most sensitive personal information a client ever shares: income, assets, debts, family circumstances, and long-term goals. AFSL obligations and the Privacy Act create hard limits on where that information can go. Public AI tools sit well outside those limits.

Common use cases

·Drafting Statements of Advice and Records of Advice
·Summarising client portfolio positions and review notes
·Generating plain-English explanations of complex financial products
·Preparing compliance documentation and file notes
·Drafting client correspondence and annual review summaries
·Answering staff queries against internal compliance procedures

Regulatory fit

Corporations Act 2001 (AFSL obligations)

Australian Financial Services Licence holders have statutory obligations to maintain client confidentiality. Disclosing client financial information to an overseas AI provider is an uncontrolled third-party disclosure.

Privacy Act 1988 (APP 8)

Client financial records contain personal information. Sending that information offshore to an AI service triggers cross-border disclosure requirements that most practices are not complying with.

ASIC regulatory guidance

ASIC expects licensees to have adequate risk management systems. Using unvetted public AI tools with client data represents an undocumented and unmanaged operational risk.

Also serving

HR and recruitment firms

Recruitment and HR consultancies handle sensitive personal data as a matter of course: employment history, performance records, health disclosures, background checks, and salary information. The Privacy Act treats much of this as sensitive information requiring a higher standard of protection than ordinary personal data.

Common use cases

·Drafting position descriptions, offer letters, and employment contracts
·Summarising candidate applications and interview notes
·Generating screening criteria and structured interview questions
·Drafting performance improvement plans and HR correspondence
·Answering staff queries against HR policies and award conditions
·Preparing client reports on talent pipeline and market conditions

Regulatory fit

Privacy Act 1988 (Sensitive Information)

Health information, union membership, and certain other employment-related data are classified as sensitive information under the Privacy Act, requiring explicit consent before collection and strict limits on disclosure.

APP 8 Cross-Border Disclosure

Candidate and employee data processed by an overseas AI provider constitutes a cross-border disclosure. The disclosing organisation remains accountable if the overseas recipient mishandles the data.

Client confidentiality obligations

Recruitment firms receive confidential workforce and commercial information from client organisations. Uploading that information to a public AI tool breaches the trust on which the client relationship depends.

Also serving

Architecture and urban planning

Architecture and planning firms work with client briefs, site analyses, and design concepts that are commercially sensitive from the first conversation. Development projects often involve confidential negotiations, landholder information, and strategic commercial positions that cannot be shared outside the firm.

Common use cases

·Drafting design rationale, planning submissions, and project reports
·Summarising council feedback, objections, and statutory requirements
·Reviewing planning instruments and development control plans
·Generating client briefing documents and project status updates
·Answering staff queries against the firm's own project archive
·Producing plain-English summaries of complex regulatory conditions

Regulatory fit

Client confidentiality and NDA obligations

Most architecture engagements involve confidential client briefs and commercially sensitive project information. Non-disclosure agreements are standard. Public AI tools represent an uncontrolled disclosure to a third party.

Privacy Act 1988

Development projects often involve personal information about landholders, objectors, and community members. Handling that information through a public AI model creates disclosure and cross-border transfer risks.

IP ownership and design confidentiality

Design concepts, drawings, and methodologies may have commercial value independent of the project. Uploading them to a public model raises questions about IP ownership and competitive exposure that are difficult to resolve after the fact.

Also serving

Management consulting

Management consultants are trusted with commercially sensitive information that clients would not share with a competitor: strategic plans, financial performance, operational weaknesses, and pending transactions. That trust is the basis of the engagement. Public AI tools break it.

Common use cases

·Drafting strategy documents, board papers, and executive presentations
·Summarising interview findings, workshop outputs, and stakeholder feedback
·Analysing operational data and generating recommendations
·Producing client deliverables and project status reports
·Answering engagement team queries against project documentation
·Generating first drafts of proposals and scope of work documents

Regulatory fit

Client confidentiality obligations

Consulting engagement agreements universally include confidentiality clauses. The obligation covers all information received from the client. Using a public AI tool with that information is a breach regardless of intent.

Privacy Act 1988

Consulting engagements regularly involve personal information about employees, customers, and stakeholders. Disclosing that information to an overseas AI provider triggers cross-border obligations most firms are not managing.

Commercial sensitivity and market risk

Merger activity, pricing strategies, and competitive positioning disclosed to a public AI model may influence model outputs for other users. The risk is difficult to quantify and impossible to remediate after disclosure.

Also serving

Insolvency and restructuring practitioners

Insolvency practitioners work with information that is sensitive on multiple dimensions at once: financial distress, creditor negotiations, employee entitlements, and pending legal action. The regulatory framework governing registered liquidators and trustees imposes strict duties that public AI tools are not designed to support.

Common use cases

·Drafting creditor reports, DIRRI declarations, and statutory notices
·Summarising financial records, trading history, and asset schedules
·Reviewing proofs of debt and generating creditor correspondence
·Preparing committee of inspection minutes and meeting materials
·Answering staff queries against the firm's own precedent library
·Generating plain-English communications for creditors and employees

Regulatory fit

Corporations Act 2001 (Insolvency obligations)

Registered liquidators and administrators have statutory duties of confidentiality to the insolvent estate, creditors, and affected parties. Disclosing estate information to a public AI provider is an uncontrolled third-party disclosure with no regulatory basis.

ARITA Code of Professional Practice

The Australian Restructuring Insolvency and Turnaround Association code requires members to maintain confidentiality of all information obtained in the course of an appointment. This obligation extends to how information is processed and by whom.

Privacy Act 1988

Insolvency estates routinely involve personal information about directors, employees, and creditors. Processing that information through an overseas AI provider creates cross-border disclosure obligations and potential liability.

Is your industry on this list?

If your firm handles confidential client information and you're thinking about AI, we should talk.

Book a consult