This Lawyer at OpenAI Uses ChatGPT Everyday

Nicole Diaz, an associate general counsel at OpenAI, represents a new category of AI user: a professional with no software engineering background who is creating custom workflows to handle policy drafting, employee disclosures and administrative tasks. Her experience offers a glimpse into the company’s broader vision for artificial intelligence, one in which lawyers, accountants, human resources managers and other knowledge workers increasingly build their own software, blurring the line between subject-matter expertise and technical development.

When Nicole Diaz joined OpenAI’s legal department last year, she had never written code. Today, the associate general counsel uses ChatGPT and Codex, the company’s artificial intelligence coding agent, to automate parts of her daily work, from drafting employee guidance to sorting compliance inquiries.

“My sense of what’s possible,” Diaz said, “has rapidly expanded in the last six months, even three months.”

Her experience illustrates a broader shift taking place inside OpenAI and across corporate America. As increasingly capable AI systems move beyond answering questions and generating text, companies are testing whether employees with little or no technical background can build custom software tools for themselves.

Diaz works in corporate compliance, the legal function responsible for ensuring that a company’s employees adhere to laws and internal policies. It is the kind of department that typically grows in importance as companies prepare for the scrutiny associated with a public listing.

For the past year, Diaz has used OpenAI’s technology to streamline that work. The tools do not replace her legal judgment, she said. Instead, they handle many of the repetitive tasks that surround it.

Diaz followed a conventional legal path before arriving at OpenAI. She studied law at Harvard, worked as a litigation associate at Skadden and later served as a compliance lawyer at Snap. None of those roles involved building software. OpenAI changed that.

Her experience reflects one of OpenAI’s central ambitions: that AI agents will enable professionals who have never considered themselves programmers to create tools tailored to their own work.

It also offers a glimpse into a debate unfolding across the legal industry. As law firms and corporate legal departments evaluate whether to adopt specialized legal software platforms such as Harvey, Diaz’s workflow suggests how far lawyers can go using frontier AI models directly.

One of the first tasks she tackled was policy writing.

Outside law firms often provide draft policies filled with legal terminology and dense language, documents that may satisfy legal requirements but are difficult for employees to navigate. Diaz previously rewrote those policies manually, translating them into practical guidance.

She now uses a custom ChatGPT skill called “simplify” to perform much of that work.

The tool shortens sentences, removes legal jargon and reformats documents to match OpenAI’s internal policy templates. Diaz still reviews every draft and sends revised policies back to outside counsel for approval, but she said the process has become significantly faster.

Employees also regularly seek advice on potential conflicts of interest, such as whether they can take advisory positions, invest in a friend’s startup or attend events hosted by government officials.

Previously, Diaz relied on a collection of draft responses stored in a Google document. She still had to determine which guidance applied, draft a response and maintain records of the outcome.

Today, Codex helps manage that process.

Each evening, an AI agent reviews her inbox, identifies conflict-related disclosures, categorizes them by risk level and drafts responses based on guidance Diaz has already approved.

Routine inquiries, such as angel investments, may receive standard replies. More complex situations trigger requests for additional information, while higher-risk matters are escalated for Diaz’s review.

The system also creates a record of incoming disclosures, tracks response times and helps identify recurring areas where employees may need clearer guidance.

The technology remains imperfect. Diaz said one of her main frustrations is that AI-generated responses can still sound overly formal.

She is developing what she describes as an “about me” file, a set of instructions that teaches ChatGPT how she writes and communicates so that responses more closely reflect her own style.

Learning these skills is not part of a traditional legal education. At OpenAI, Diaz said, much of the training happens informally.

There is no expectation that every lawyer becomes a heavy user of Codex. Diaz credits a colleague, Bright Kellogg, with helping lawyers adopt the tools gradually by sharing practical examples and highlighting projects built by others on the team.

For Diaz, that exchange of ideas has become one of the most rewarding aspects of the experience. Discovering new tools built by colleagues can feel, she said, like trading Pokémon cards, with one person showing off a skill with “really cool powers” that others immediately want to try.

Stay ahead of the stories shaping our world. Subscribe to Impact Newswire for timely, curated insights on global tech, business, and innovation all in one place.

Dive deeper into the future with the Cause Effect 4.0 Podcast, where we explore the ideas, trends, and technologies driving the global AI conversation.

Got a story to share? Pitch it to us at info@impactnews-wire.com and reach the right audience worldwide