AI Is in Your Codebase: What Every IT Leader Needs to Know (Before the Board Asks)

Your non-human Engineering colleagues are here. Marco Argenti, chief information officer of Goldman Sachs, recently said that they have introduced in to their workforce ‘Devin’, which has been dubbed as the first AI Software Engineer. “We’re going to start augmenting our workforce with Devin, which is going to be like our new employee who’s going to start doing stuff on the behalf of our developers,” Argenti told CNBC.

🔍 Why This Matters

AI in software development is no longer theoretical or experimental — it’s embedded in how your organization builds, tests, and ships products. It might soon show up, as is the case with Goldman Sachs in your organization as the next employee you hire . And whether you’ve formally approved it or not, your teams and your vendors are already using it in some form or shape.

The board will ask, “What’s our AI strategy?”

From Experiments to Infrastructure

According to McKinsey’s 2024 State of AI report:

65% of companies now use generative AI in at least one business function.
In software development, adoption jumped from 23% in 2023 to 78% in 2024.

This is no longer about pilot projects or hackathons.

AI is now:

• Embedded in your IDEs
• Suggesting your team’s code
• Generating test scripts
• Recommending cloud configurations
• Writing documentation — before your developers do
• Is the next developer you ‘hire’

If you lead IT or software delivery, AI is already part of your landscape, whether you planned for it or not.

📘 Lessons from Google’s AI Engineering Playbook

Google has quietly led the way in integrating AI into engineering. Their internal research surfaces a few critical lessons for all IT leaders:

✅ Seamless Integration Is Everything

“The most successful AI tools were those that blended into the workflow — not those that required users to change behavior.”

If a developer has to stop and remember to use AI, it won’t scale.

📊 Real Metrics Come From Real Usage

Offline benchmarks don’t reflect true value. Google uses live A/B testing to measure impact — not just prediction accuracy.

AI adoption only works when it changes real behavior, in real work.

---

What’s Hype — and What’s Real

The market is saturated with claims. Every vendor seems to offer “AI-enabled everything,” from development to delivery. And while the innovation is real, it’s important to separate signal from noise:

• ✅ Real Value: Coding assistants (e.g., GitHub Copilot), test automation, agentic workflows, and productivity analytics.
• ⚠️ Watch Closely: Generative code translation, complex LLM integrations, and “AI-first” platforms not built for enterprise scale.
• ❌ Caution: Black-box models with limited explainability or tools that bypass compliance and security guardrails.

AI in software engineering is real. But value depends on fit-for-purpose use cases, clean integration, and disciplined governance — not just vendor demos.

🛠️ What IT Leaders Must Do Now

1. Watch How Developers Use AI

Spend time with your teams. Notice what tools they trust, what they ignore, and where they’re overriding AI. Those friction points? That’s your roadmap.

2. Set Guardrails, Not Just Guidelines

AI can write insecure code. It can hallucinate. Build automated review checks into your SDLC — security scans, licensing validation, model limits. These have to extend and apply to your entire IT ecosystem including the 3rd party partners you work with.

3. Upskill Across the Stack

This isn’t just about hiring data scientists. Developers, QA, DevOps, and architects all need to learn:

• Prompting best practices
• Critical review of AI outputs
• Risk mitigation with AI-generated content

4. Treat Data Like Product Infrastructure

If you want valuable AI, you need clean, tagged, enterprise data. Invest in metadata pipelines, feedback loops, and internal model telemetry.

---

🧭 Final Thought: Lead with Calm, Not Hype

AI has arrived. Quietly, steadily, and permanently.

This isn’t a wave to ride — it’s a platform shift to architect around.
And in the noise of pilot projects, platform launches, and boardroom buzz, your role is simple: bring clarity.

Be the one who knows:

• What’s real
• What’s risky
• And what’s ready

Because the next time your CEO says, “Tell me what we’re doing with AI,”
you won’t need to ask anyone else — you’ll already have the answer.