AI at Gilbert + Tobin: A Law Firm CEO's $62 Annual Review

A confidential annual review process for 120 partners cost Australian law firm Gilbert + Tobin about $62 in AI compute — work that would otherwise have required several people over multiple days. That figure comes from Sam Nickless, Partner and Chief Executive Officer at Gilbert + Tobin, one of Australia's leading corporate law firms, and it captures why the firm has put AI at the centre of how it now runs its business.

The story matters beyond one firm. Law is a profession built on specialist knowledge, careful judgement and the secure handling of sensitive information — exactly the qualities sceptics once said generative AI couldn't touch. Gilbert + Tobin's experience shows how a traditional partnership is turning that assumption around, with governance, security controls and a CEO who uses the tools himself.

What did the firm actually deploy?

Gilbert + Tobin moved through three deliberate stages. It began by giving its people room to experiment with AI, with one hard rule: no client data or confidential information should be entered. That created what Nickless calls "a safe, constrained environment to experiment."

The second stage brought a secure channel to powerful AI. The firm introduced ChatGPT Enterprise in a controlled environment and expanded from there, later adding ChatGPT Work and Codex. The goal was to move from individual assistance to defined, multistep agentic workflows — with people responsible for review and approval at every step.

The security case rested on concrete features, not trust. Australian data residency, contractual protections, role-based access and administrative controls, Nickless said, "gave us the confidence to scale while meeting our requirements and those of our clients." Takeaway: the firm scaled AI only after the compliance plumbing was in place.

Why does the CEO say "AI is not cheating"?

Nickless frames governance as an enabler, not a brake. "Our message was simple: AI is not cheating. It is another tool people can use to apply their judgement," he said. "They still need to define the task, check the output and take responsibility for the final work."

That framing sets responsibility where a law firm needs it — with the lawyer, not the tool. The model handles drafting and assembly; the human defines the task, checks the result and signs off on the final work product.

How much did the annual review actually save?

The most concrete number in the story comes from the firm's confidential annual review process, which involves all 120 partners. Nickless used ChatGPT Work to update files with data from a spreadsheet, create PDFs and bundle the finished material.

"It completed the work in minutes," he said. When he checked the usage dashboard afterwards — ChatGPT Work now provides detailed, granular data on usage — the cost came to about $62. In his words, "That was a good trade for the time I saved – and for work that would otherwise have required several other people over multiple days."

For a firm where partner time bills at corporate-law rates, a $62 AI bill for a multi-day task is the kind of arithmetic that explains why adoption is accelerating across professional services. The lesson is that granular cost visibility makes the value case undeniable.

What is the CEO's "digital twin"?

Nickless didn't stop at delegating tasks to AI — he built a version of himself. He created a "digital twin" through a custom GPT (now shifting to Plugins), trained on examples of his writing, feedback he had received, context about the firm and the way he likes to work.

It started as a personal workload tool and became more valuable when he shared it with his direct reports. "My team is now able to use it to get 'my feedback', effectively, via ChatGPT," he explained. "It lets them test their thinking against my context and working style before coming to me."

The twin has clear limits, which Nickless is careful to state: "It does not make decisions or speak for me; it helps our conversations start from a stronger place." Takeaway: the digital twin compresses routine feedback loops, while genuine decisions stay with the human CEO.

Did the firm set usage mandates?

No — and that choice is deliberate. "We did not mandate usage targets," Nickless said. "Instead, we made the technology relevant."

The firm's business transformation team demonstrated workflows tailored to different functions. Videos, showcases, challenges and an early AI bounty surfaced ideas from across the firm. Nickless points to the breadth of participation as the real win: "What has been most encouraging is the creativity coming from every part of the organisation – not only the technology team."

Why does the CEO say leaders must use AI themselves?

Nickless's advice to other chief executives starts with personal use. "Start by using the technology yourself," he said. "Experiment with real work, understand its limitations and show your people that learning matters."

Visibility from the top changes behaviour. "When people see that AI matters to you, it sets the tone and encourages others to engage with it," he said, adding: "Getting started yourself…sets the tone and encourages others."

He pairs that leadership with two organisational requirements: clear governance and practical enablement. Give teams examples tailored to their roles, create safe ways to test ideas and keep people responsible for the result.

Is the value really about cutting jobs?

Nickless is explicit that it isn't. The firm has three objectives, he said: "to use AI to deliver better services to our clients, make Gilbert + Tobin a great place to work for our people and firmly establish ourselves as a leader in our market."

The measured value so far is qualitative. "The real value we've seen from AI so far hasn't been about reducing cost or needing fewer people, but freeing our people to focus on the more human side of their work, meeting with clients, coaching their teams and mentoring the future leaders of our industry," he said. "It's been about…freeing them up to then do the more human side of their work."

That claim carries weight in a market where legal hiring pressure remains high and firms compete on client relationships rather than raw document throughput. Gilbert + Tobin has always been willing to lean into new technology, and the firm's bet now is that AI expands what its lawyers can produce rather than shrinking the team that produces it.

What's the bigger picture for law firms?

Nickless closes with a scope claim that goes well beyond efficiency. "The opportunity is bigger than doing the same work a little more quickly," he said. "Used well, AI can raise the quality and volume of what an organisation produces – and give people more time for the complex, strategic work where their expertise and judgement create the greatest value."

For an industry whose billable-hour economics are built on scarce expert time, that is the real disruption: raising output quality and volume simultaneously, rather than trimming minutes off tasks. Gilbert + Tobin's agentic workflows — reviewed and approved by humans — offer one template for how a 120-partner firm gets there without betting its confidentiality obligations on it.