Few IT executives feel the pace of developments in artificial intelligence (AI) as acutely as Manu Narayan, the first chief information officer at GitLab. Nine months into his role at the software development platform – which now exceeds $1 billion in revenue and employs more than 2,000 people – Narayan is tasked with turning the company into a proving ground for the very technologies its customers use.

The AI landscape is shifting so quickly that GitLab has repeatedly revisited its goals. Narayan's mandate is mostly internal: modernizing the business application stack, user support, and data and analytics. But instead of bolting AI onto existing workflows, his goal is to rebuild operations from the ground up. When revisiting the AI strategy a few months ago, the focus was not on how to introduce AI, but on rethinking the nature of work internally, leveraging AI. The approach involves thinking about processes from first principles and then using agentic AI to drive them.

The Problem with Tokenmaxxing

As AI adoption increases across enterprises, CIOs naturally grapple with cost control and measurement. However, Narayan is wary of strategies such as “tokenmaxxing”, where developers and employees are encouraged to maximize the number of AI tokens they use. He argues that gamification can drive outcomes, but often drives incorrect behavior. Measuring success purely by context-in and context-out is problematic because it's hard to know if someone is gaming the system – sending excessive content just to inflate token counts.

Instead of tracking token burn, GitLab tracks daily active usage across its tech stack to ensure the workforce is building sustainable habits. For calculating hard return on investment, Narayan insists on anchoring AI deployments to traditional business metrics. For example, for an AI agent assisting a sales development representative, success isn't measured by the number of prompts generated, but by standard key performance indicators: outbound messages, meetings scheduled, and sales pipeline conversion.

Building with Agentic AI

Pointing to a customer success manager (CSM) as an example, Narayan noted that the purpose of the role is to build deep client relationships, yet CSMs spend hours on administrative tasks such as building quarterly business review slides for clients, transcribing notes, and hunting for context across customer relationship management systems, data warehouses, and chat channels. By deploying AI agents to handle that grunt work, GitLab is looking to free up its workforce to focus on high-level strategy. The company wants all team members to focus on what matters most: the core purpose of their role. AI is used for tasks that can help them scale out in a more linear way, not just a 10-15% increase in productivity.

To manage these AI deployments, GitLab has adopted a hub-and-spoke operating model. A central AI enterprise team handles governance, technical building, and guardrails, while dedicated “AI transformation owners” embedded in individual divisions identify time-consuming, repeatable work that is ripe for automation. The approach has already been applied to GitLab's own internal employee support network. The company has built AI agents to assist its 120 internal support staff across IT, people operations, and sales, helping them instantly pull the context they need or deflect routine tickets entirely.

Build vs. Buy and the Future of SaaS

As AI lowers the barrier to building internal tools, some have suggested that off-the-shelf software-as-a-service (SaaS) applications are numbered. Narayan views this as vastly overstated, particularly from a governance and compliance perspective. He believes we may see more custom interfaces and the disaggregation of systems of interaction from systems of record, but the underlying governance controls in core SaaS tools aren't going anywhere.

Narayan also pointed to the hidden costs of bespoke software development. It's easy to get to 90% of an application developed in-house, but the last 10% – role-based access controls, auditability, immutable logging, which are things needed as a public company or a company that deals with regulated customers – is incredibly complex to build. To ensure safety across custom and supplier tools, GitLab grounds its AI governance in a strict data classification standard. Public data flows through self-service platforms, while proprietary or customer data requires deeper security reviews before interacting with large language models.

Change Management and the Clock is Ticking

Despite strong executive backing and budget, change management remains a challenge. Bridging the gap between AI-forward employees and those who are slower to adapt requires a mix of departmental centers of excellence and internal AI hackathons. Yet, for a CIO, the greatest pressure is the ticking clock. Narayan worries whether they are moving fast enough. In the AI era, decision-making needs to happen in days and weeks, not months and quarters. But he still worries about whether the right initiatives are being driven to deliver long-term ROI.

GitLab's approach mirrors broader industry trends. Many enterprises are moving away from simple chatbot-style AI toward agentic systems that can autonomously execute multi-step tasks. This shift requires not only new technology but also organizational redesign. The hub-and-spoke model GitLab uses is similar to that adopted by other large tech firms, where a central AI team sets policy and provides reusable components, while distributed teams apply those tools to their specific domains. This structure helps avoid fragmented AI efforts while still allowing flexibility.

Narayan's background brings credibility to this transformation. He previously held senior IT roles at companies like Autodesk, where he led digital transformation initiatives. His experience includes scaling cloud infrastructure and implementing enterprise resource planning systems. At GitLab, he is applying those lessons to an AI-first world. The company itself is known for its all-remote workforce and DevOps platform, making it a natural lab for AI-driven productivity gains.

The rejection of tokenmaxxing is notable because many organizations still use consumption-based AI pricing models. By focusing on business outcomes rather than token volume, GitLab avoids the pitfalls of inflated usage without value. This aligns with best practices from industry analysts who caution against rewarding AI usage for its own sake. Instead, Narayan advocates for measuring AI's impact on core business metrics, such as customer retention, revenue growth, or employee satisfaction.

Looking ahead, GitLab plans to expand its AI agent deployment across more departments. The company is also exploring how AI can help with software development itself – for example, by automating code reviews or generating documentation. Since GitLab's own product is a DevSecOps platform, these internal experiments feed directly into product improvements for its customers.

Source: ComputerWeekly.com News