How Australian Law Firm Gilbert + Tobin Scaled OpenAI Tools Across Its Practice

To move beyond informal testing, the firm adopted ChatGPT Enterprise alongside ChatGPT Work and Codex within a managed environment. Meeting corporate and client security standards required specific technical safeguards, including Australian data residency, contractual protections, administrative controls, and role-based access. Firm leadership established a clear governance principle: AI is treated as a productivity tool rather than a shortcut, meaning practitioners must frame prompts, verify outputs, and retain formal accountability for all completed deliverables.

Adoption was supported through internal engagement rather than mandatory usage quotas. The firm's business transformation group showcased tailored workflows across distinct practice groups, while internal challenges, instructional videos, and an early AI bounty surfaced ideas from staff across legal and non-technical departments.

Practical applications: From partner reviews to digital twins

Nickless integrated the technology into his own daily responsibilities, building a custom GPT—now transitioning toward Plugins—that functions as an internal digital twin. Configured using samples of his writing, past feedback, and operational context about the firm, the model allows direct reports to test concepts against his working style before scheduling formal meetings. Nickless stresses that the system does not make executive decisions or speak on his behalf, but rather ensures collaborative discussions begin from a clearer baseline.

In another instance, Nickless deployed ChatGPT Work to manage a confidential annual review workflow involving 120 partners. The system processed spreadsheet records, modified files, generated PDFs, and compiled the documentation within minutes. A subsequent review of the administrative usage dashboard showed the task cost approximately $62 in compute, replacing a process that traditionally required multiple staff members working over several days.

What it means for developers

Gilbert + Tobin’s experience highlights critical considerations for developers building AI software for regulated professional sectors. First, strict governance and infrastructure compliance—such as regional data residency and verifiable access permissions—are non-negotiable requirements before enterprise legal teams will deploy models to production. Enterprise systems require transparent administrative reporting, as demonstrated by the usage dashboard that allowed leadership to verify the cost and efficiency of multi-file data operations.

Second, the firm's progression shows that enterprise software is shifting from simple text generation toward multistep agentic workflows. When designing these systems, developers need to implement structured checkpoints where human operators review and sign off on outputs, ensuring the final legal or analytical responsibility remains human-led.

For engineers testing and benchmarking such workflows across different architectures, evaluating cost-to-performance tradeoffs is essential. Developers looking to experiment with multiple foundation models can access platforms like Apixo (https://apixoai.online), which offers cheap pay-per-token API access to GPT, Claude, Gemini, Grok, and DeepSeek under a single key. As demonstrated by the firm's $62 administrative run, cost-efficient token pricing makes large-scale document compilation and batch processing viable for complex corporate environments.

Source: Gilbert + Tobin rethinks the business of law with AI — OpenAI News. Written by the Apixo team from that report.