Editorial illustration for EY boosts coding output 4‑5× by linking AI agents to engineering standards
EY Multiplies Coding Output 4-5x with AI Agent Strategy
EY boosts coding output 4‑5× by linking AI agents to engineering standards
Four times, sometimes five times faster. Not a headline from a lab experiment, but a real-world result across EY’s sprawling suite of audit, tax, and financial platforms. This wasn’t about plugging in a tool and watching the magic happen.
For 18 to 24 months, EY’s engineering teams laid groundwork, cultural and technical, before their AI agents could run. They started small, with Copilot-style assistants, letting developers get comfortable with prompt engineering on their own terms. Organic adoption, not a leadership mandate.
The real unlock came when Newman’s team realized agents were useless without context: EY’s code repos, engineering standards, and source catalogs. No context universe, and you get generic output that needs a full rewrite. So they pitted three platforms, Lovable, Replit, and Factory’s Droids, against each other, tracking adoption and productivity across every team.
No mandates, just results.
The result: 4x to 5x productivity gains across teams building EY's suite of audit, tax, and financial platforms. But the gains didn't come from just turning on a tool. Newman's team spent 18 to 24 months building the cultural foundation and technical integrations that made semi-autonomous coding work at scale.
EY started with GitHub Copilot-style tools, letting engineers get comfortable with prompt engineering and assistive AI. Newman said the key learning was making AI adoption organic rather than forced from leadership. "It's important to bring AI capabilities as a ground-up organic adoption rather than force them onto the users," he said.
Developers wanted to move beyond code generation to building, deployment, and operationalization. Newman realized agents needed access to EY's code repos, engineering standards and source catalogs to generate deployable code. Without that "context universe," as Newman calls it, agents produce generic output that requires extensive rework.
EY evaluated multiple agent platforms: Lovable, Replit and Factory's IDE-based Droids. Rather than mandate a tool, Newman's team measured adoption, usage and productivity across all three.
The lesson from EY is not about the tool. It’s about the soil. You can plant the most powerful AI agent into barren ground, and it will wither.
Newman’s team spent two years cultivating a culture where engineers *wanted* to adopt, not because they were told to, but because the tools actually made their work better. They built the technical infrastructure, the code repositories, the standards. They gave the agents context.
And then they let the teams choose their own instruments. The result is a four- to fivefold leap, not a forced march, but an organic growth. That is the real blueprint for scale: prepare the ground, then let the technology take root.
Related Reading
Policy & Regulation Grammarly faces class-action suit over AI ‘Expert Review’ feature
Policy & Regulation Claude Mythos highlights EU AI safety gaps, says researcher Caroli
Policy & Regulation After ditching AI fitness apps and a Fitbit, I return to Peloton classes
From Archives Google sued over Gemini chatbot allegedly coaching 36‑year‑old to suicide
From Archives Pentagon embeds Claude, sole cleared AI, into classified tech amid culture wars
Common Questions Answered
How did EY achieve 4-5x productivity gains in coding?
EY linked AI coding agents to internal engineering standards over 18-24 months, carefully integrating the technology into their workflow. The approach involved starting with GitHub Copilot-style tools, allowing engineers to gradually become comfortable with AI-assisted coding and developing a robust cultural and technical foundation.
What challenges did EY encounter when implementing AI coding agents?
Despite generating thousands of lines of code quickly, EY found that most AI-generated code failed compliance checks or breached coding standards. Stephen Newman emphasized that generating code does not automatically mean the code is usable or valuable, highlighting the need for careful integration and validation.
What was EY's strategy for introducing AI into their coding process?
EY took a methodical approach by first introducing GitHub Copilot-style tools to help engineers get comfortable with prompt engineering and assistive AI. They spent 18-24 months building a cultural foundation and technical integrations to make semi-autonomous coding work effectively across their audit, tax, and financial platforms teams.
