Gilbert + Tobin’s leadership treats generative AI very differently depending on the task at hand, steering lawyers away from hallucination-prone tools in legal work but leaning heavily on them for practice management. OpenAI has identified the law firm as one of its most engaged Australian clients, a label that reflects how deeply ChatGPT has been woven into the firm’s internal workflows.
Core to this shift is OpenAI’s Codex autonomous agent, which the firm now uses to strip manual labour out of previously tedious and repetitive processes. For a partnership that includes 112 partners, that kind of automation changes how administrative work gets done.
Codex has already transformed one of Gilbert + Tobin’s more painful internal rituals, updating and distributing confidential appraisal documents for every partner. Each partner submits sensitive paperwork, which previously demanded painstaking manual edits to add updated figures, export files and collate them.
Codex now takes those June 30 numbers, inserts them into every document, re-exports each file as a PDF and gathers them into a single folder, before packaging and sending the bundle to the relevant committee. The entire workflow, which once consumed significant staff hours, reportedly now takes about 10 minutes end to end.
Quiet productivity wins like that are driving OpenAI to push ChatGPT and Codex into enterprise settings, as rivals like Anthropic focus attention on cutting-edge scientific capabilities. Claude attracts interest inside firms like Gilbert + Tobin, and some lawyers lobby to try it, especially when benchmarks showcase breakthroughs on complex scientific problems.
Leadership judges success on a different metric, how well these systems handle routine but high-stakes tasks like spreadsheet updates and document routing without compromising confidentiality. In that race, OpenAI’s tools look less like experimental lab gear and more like the new back office infrastructure for knowledge-heavy businesses.
