Most engineering leaders assume that if AI helps developers write code faster, software gets to production faster too. ClearRoute CEO, James Jarvis, says the data tells a different story and understanding why, matters for anyone setting AI investment priorities this year. 

The Real Bottleneck Was Never Coding Speed

ClearRoute's State of the Route to Live Report 2026 research found that AI is accelerating development work without shortening the path to live software. Jarvis explains the reasoning plainly: producing more code faster does nothing for a business until that code is actually deployed and generating value. Until then, it sits as a cost rather than a return.

The constraint sits downstream of coding, in the layers of oversight that decide whether something is safe and ready to launch, including security checks, approval steps and testing standards. A newer, leaner company can ship an idea within days. Larger, more established organisations often take months for the same step. Speeding up the coding stage without addressing that gap simply pushes the pressure further down the pipeline rather than relieving it.

AI Is an Amplifier, Not a Fix

Jarvis frames AI as an amplifier of whatever operating model already exists. Organisations with strong engineering discipline see AI compound their advantage. Organisations with slow or fragmented release processes see AI compound that friction instead. This reframes the leadership question from "how do we get more AI" to "what state is our delivery pipeline in before we add more throughput to it."

What Enterprise Leaders Are Actually Wrestling With

Drawing from his regular conversations with C-level executives at large, regulated organisations, Jarvis identifies four recurring challenges:

• Selecting and prioritising the right AI use cases, and defining how success will be measured

• Building confidence that the expected return justifies the investment, with the bar set at returns that clearly outweigh the cost

• Moving AI work beyond pilots so it becomes a standing part of how the wider organisation runs, not a one-off trial

• Supporting employees through the shift in culture and working habits that follows once AI takes on tasks they used to do

He notes that at scale, this becomes as much a conversation about culture and change as it is about technology.

Becoming AI Native Requires Structure, Not Just Access

ClearRoute's own shift to an AI native operating model started with building an internal platform on foundational primitives, led by Global Head of Engineering and AI, Justin Wilkin. From there, every employee, technical and non-technical, received access to build internal skills and agents. Today, agents support parts of HR, talent acquisition, finance, operations and sales, all coordinated through Orbit, ClearRoute's platform for running and overseeing agents, with governance, access controls and cost tracking built in from the start.

ClearRoute also placed more than 50 AI-focused architects directly with clients to identify real workflows and turn them into standing, well-managed systems the business can build on, rather than isolated builds that live on one person's laptop. The distinction Jarvis draws is between AI experimentation and AI infrastructure. Only the latter compounds.

Leadership Habits Are Changing Too

Jarvis applies the same approach to his own workflow. An agent monitors Salesforce and surfaces pipeline changes without requiring him to log in. On calls, an agent transcribes, logs activity and drafts follow up emails in his tone for review. Board reporting now pulls from MCP connected systems into a branded pack with analysis and recommendations, which the team validates before use. In talent acquisition, agents screen incoming applications against an ideal candidate profile and produce a weekly conversion report by channel, freeing a small team to focus where their judgement adds the most value.

Key Takeaways

• Coding speed and delivery speed are different problems. Faster development does not shorten the path to live software if downstream governance and release processes are not addressed first.

• AI amplifies existing operating maturity. Strong engineering discipline compounds with AI. Weak processes compound too, in the wrong direction.

• Use case prioritisation and ROI clarity remain the top blockers for enterprise leaders moving from pilot to scale.

• Cultural change is inseparable from AI adoption. Freeing up employee time only creates value if organisations plan for what comes next.

• Governance and cost visibility are prerequisites for scale, not afterthoughts, as shown by ClearRoute's own platform build.

• Embedded expertise turns AI from a one-off build into a durable capability. Client-facing AI architects convert ad hoc automation into well-managed, reusable systems.

Explore the full State of the Route to Live 2026 report for the complete data behind this gap.

Read the full interview with James Jarvis on TechInformed for more on his approach to AI native leadership.