Gilbert + Tobin has transitioned from cautious experiments to firm-wide, governed AI workflows that the Australian corporate law firm says are reshaping how it delivers client work and organises internal tasks. The firm combined early, constrained testing with controlled rollouts of ChatGPT Enterprise, ChatGPT Work and Codex to move beyond one-off assistance into multistep, agentic processes — while keeping lawyers ultimately responsible for review and approval.
Sam Nickless, a partner and the firm’s chief executive officer, described a deliberate three-phase approach: experimentation, personal productivity and organisational productivity. In the first phase, staff were encouraged to explore ChatGPT but restricted from entering client or confidential information. That initial, conservative stance let practitioners assess capabilities safely before a secure channel was introduced for broader use.
To scale, Gilbert + Tobin deployed ChatGPT Enterprise and later added ChatGPT Work and Codex, combining Australian data residency, contractual protections, role-based access and administrative controls to meet internal and client requirements. Nickless emphasised that governance was designed to enable responsible use: lawyers still define tasks, check outputs and remain accountable for final deliverables rather than ceding control to the tools.
Adoption across the firm has been striking. Among users enabled for ChatGPT the firm reports 87% active usage — more than double typical uptake for other enterprise tools. Nickless credited that engagement to leadership modelling and to relevance: instead of mandating usage, leaders used the technology themselves and the firm worked to make AI demonstrably useful across roles.
Practical enablement came from tailored workflows and programmes run by the business transformation team. Gilbert + Tobin produced videos, showcases and internal challenges and even launched an early AI bounty to surface practical ideas. That approach sought creativity from across the organisation, not only from technologists, and aimed to show how AI could be applied to everyday legal and administrative tasks.
Nickless gave several concrete examples of the firm’s workstreams. He created a custom GPT described as a “digital twin” that captures examples of his writing, feedback, firm context and preferred working style. Shared with his direct reports, the custom assistant lets them test ideas and get preliminary feedback in his voice before issues are escalated to him; it does not make decisions or represent him in authoritative ways.
For confidential processes, the firm used ChatGPT Work. Nickless ran a confidential annual review workflow for the firm’s 120 partners, asking the product to update files with spreadsheet data, generate PDFs and bundle the materials. The task completed in minutes, and after reviewing usage dashboards and admin controls he estimated the direct cost of that automation at about $62 — a sum he judged a worthwhile trade-off for time saved and for work that otherwise would have required several people over multiple days.
Throughout the rollout, Gilbert + Tobin maintained a strong emphasis on governance and controls. Role-based access, data residency and contractual protections were implemented to align AI use with client confidentiality and regulatory expectations. Nickless consistently framed AI as an additional tool for professional judgement and reminded staff that outputs must be reviewed: users define tasks, inspect outputs and take responsibility for final work.
His advice to other CEOs was practical and unequivocal: start by using the technology yourself, experiment with real work to understand its limitations, and show that leadership engagement matters. He urged combining that hands-on leadership with clear governance and targeted enablement: give teams role-specific examples, create safe spaces to test ideas and keep people accountable for outcomes.
On strategic impact, Nickless said the main benefit observed so far has not been headcount reduction but freeing people to focus on higher-value, human-centred activities such as client engagement, coaching and mentoring. In his view, AI has helped “free them up to then do the more human side of their work.” Used well, he added, AI can both raise the quality and increase the volume of an organisation’s output.
The firm’s experience with ChatGPT Enterprise, ChatGPT Work and Codex — and its emphasis on leadership-led adoption, governance and role-specific enablement — offers a pragmatic model for professional services firms looking to scale generative AI while preserving accountability. Gilbert + Tobin has also documented its collaboration with OpenAI in an accompanying case study on OpenAI’s site for readers seeking more detail.
As the legal sector watches, Gilbert + Tobin’s rollout underscores that integrating AI at scale is as much about people, controls and process design as it is about choosing technology. For the firm, the combination of executive sponsorship, controlled experimentation and practical workflows has turned experimental tools into productive, overseen capabilities that support lawyers rather than replace their judgement.
Source: Read the original source
Leave a Reply