News · Endava routes Codex to the front of the delivery pipeline, not just the code editor

Jul, 134 min to read
Automation

Endava routes Codex to the front of the delivery pipeline, not just the code editor

A software contractor is using OpenAI's Codex for requirements, design, and client conversations — treating the agent as a lifecycle tool rather than a coding assistant.

The agent starts before the first line of code

Most Codex stories are about writing and reviewing code. Endava's is about what happens upstream of that. The global software contractor says it uses Codex for "requirements analysis, design, specifications, development, and operations," and frames it explicitly as "a general desktop agent across our whole lifecycle" rather than a coding assistant.

That distinction is the whole point of the announcement. Endava's advice to teams starting out is to pick a non-coding workflow first — requirements analysis, design documentation, or client communication — because, in their words, the fastest way to see the tool's full value is to use it somewhere your team has never used a coding tool before.

Each of these stages used to take days or weeks of analysis. Now with Codex packaging together analysis, design, and build, we can do that as a single unified tool.Montana Labs

Diagrams as a client-facing move

The frontend of a consulting engagement is the client meeting itself, and that is where Endava reports the second visible change. Teams now produce design documents, diagrams, and specifications live during client sessions to illustrate ideas as they discuss them.

You can tell it to draw a diagram of the proposed software architecture so it's easier to understand for our clients. It rapidly accelerates the back-and-forth, and it really opens a lot of doors.Montana Labs

This is a specific, checkable behavior: the artifact that used to be produced after the meeting is now produced inside it. For a contracting firm, compressing the ideation loop with a client is a commercial edge, not just an engineering convenience.

"Codify your seniors" as an operating model

Endava's stated top leadership lesson is to capture senior architects' judgment in Codex so that junior team members receive senior guidance while they execute. Global SVP Mike Krolnik describes encoding a point of view into the tool so juniors can ask questions and produce "senior, mature-level outputs." One senior's perspective, they claim, can guide multiple less-seasoned teams in parallel.

It's worth reading this carefully. The company is describing mentorship — normally delivered through pairing and code review — being partially routed through an agent. That is a real reorganization of how expertise moves through a team, and also a bet that a senior's judgment survives being flattened into instructions the agent applies at scale.

The specific implication: the requirements gap is the target, not the keyboard

The reusable idea from Endava isn't that Codex writes code faster. It's where they pointed it. The most expensive delay in delivery work is the ambiguity between what a stakeholder wants and what engineers can build against — the phase that used to eat weeks of meetings. Endava aimed the agent at that gap first.

For applied teams evaluating agentic tools, the transferable test is narrow: can the agent turn a recorded conversation into a spec you'd actually build from, and can it render an idea as a diagram while the client is still in the room? Those are the two workflows Endava says moved the needle, and both live at the front of the project, not in the editor.

Find this story relevant to you?

Contact us to find a unique solution

Contact us

Need an AI engineering partner that can actually build?

We help businesses integrate AI, build AI-powered products, automate high-value workflows, and modernize the software systems behind them.

Get in touch

Related reading

More analysis around product delivery, operational AI, and the systems work that makes deployment hold up in reality.

Jul, 134 min to read
Automation

OpenAI reframes adoption as a 'capability overhang' problem

Jul, 134 min to read
Automation

Cisco built the majority of its AI Defense product with Codex writing the code

Jul, 134 min to read
Automation

Commonwealth Bank standardizes on ChatGPT Enterprise as the shared surface for 50,000 employees