This Law Firm Cut Hiring Tasks From 4 Hours to 20 Minutes With ChatGPT and Codex

Hi, it's Shii-chan! Today I found a news story with a bit of a different flavor — not a tech company, but a law firm putting ChatGPT and Codex to serious use. Reading about it got me genuinely excited!

OpenAI News openai.com

What was announced?

OpenAI's News page shared a case study about Gilbert + Tobin, a major Australian law firm, rolling out ChatGPT Enterprise and Codex across the whole business to speed up work and raise its quality. Gilbert + Tobin specializes in capital markets, M&A, disputes, and technology.

Rather than using AI for legal advice itself, the firm focused on improving the operations behind it — hiring, marketing, finance, and IT. They piloted ChatGPT Enterprise with the operations team first, then expanded firmwide once it proved useful. From there they rolled out Codex too, moving from single-task help toward automating multi-step workflows.

Why it matters

Law firms handle a lot of highly sensitive information and regulatory requirements, so it's not enough to adopt a tool just because it's convenient — you need real governance in place first.

Under CIO Mitch Owens, Gilbert + Tobin built out:

• Guidelines on which tasks AI can help with and what data is safe to input
• Rules requiring people to review every output
• An evaluation of how far contractual protections extend
• Role-based access control that limits what each person can see
• Data residency within Australia to build trust

It's a solid playbook for rolling out AI responsibly in a regulated industry, which is exactly why I wanted to cover it under "why it matters."

What changes

Leadership using the tools themselves seems to have shifted the whole culture. CEO Sam Nickless pushed adoption with a simple message: AI isn't cheating.

"AI isn't cheating — it's another way to apply our people's judgment and improve how we run the firm."

Instead of forcing usage quotas, the firm made people accountable for outcomes instead. That approach helped push active usage to 87% as of June 2026 — more than double the average for other tools at the firm.

Executives even have a custom GPT built as a kind of "digital twin," trained on the CEO's writing style, priorities, background, and feedback. CMO Daniel Quinn says it lets people test messaging with the custom GPT before taking it straight to the CEO, making it easier to refine ideas first (though it has no actual decision-making power).

Dive Deep

The case study shared plenty of concrete numbers.

• Hiring: Extracting candidate data dropped from about 4 hours to 20 minutes. Reference-check processing got 25 minutes faster per candidate
• Marketing and business development: Speeds up and improves the roughly 400-500 pitch decks the firm produces each year, and supports industry report writing
• Finance and tech teams: Used for spreadsheet analysis and drafting documentation and training materials

Codex's results go even further. Aviva Leitch, Head of Business Transformation, described it as turning AI "from a helper into a doer."

• Building an audit report covering 300 entities saved a full day of manual work
• Checking and renaming 1,100 files, cutting the job down by several days
• Built workflows for compliance checks — conflict checks, anti-money-laundering, politically-exposed-person screening, and identity verification — cutting tasks that used to take up to 8 hours down to minutes
• The DevOps team used Codex to build "Watchtower," a system that monitors the firm's AWS environment, automating diagnosis, improvement suggestions, and escalation

Worth noting: for actual legal work, the firm still uses accredited platforms like Harvey rather than ChatGPT.

Wrap-up

• Gilbert + Tobin rolled out ChatGPT Enterprise and Codex firmwide in stages, reaching 87% active usage
• Hiring tasks went from 4 hours to 20 minutes, audit reports saved a full day, and compliance checks dropped from up to 8 hours to minutes
• Governance — guidelines, access control, and data residency — came first, before expanding how the tools were used
• A great read for companies in regulated industries looking to adopt AI responsibly, or leaders looking for a role model for driving adoption internally