AI is growing up fast. This means we now need to both quantify and qualify the depth, breadth and scope of the AI components now being grafted onto (let’s be realistic, AI-native hasn’t happened yet) existing enterprise software structures and data management estates. Coder thinks it can help.
The self-hosted AI development infrastructure company has created an AI maturity self-assessment service. It comes with an AI ‘maturity curve’ tool, which is designed to help organizations evaluate how effectively they are adopting AI within enterprise software development.
All this comes at what may be an opportune moment, i.e., now is the time when software application development and data science teams (and the operations functions which dovetail with them) are moving from ad hoc AI usage to more structured, governed and scalable agent-driven workflows.
Outpacing Policy, Security & Platform Control
As AI tools and agents become embedded across the software development life cycle, adoption inside engineering organizations is accelerating, but often without consistent oversight. Despite pressure to accelerate AI adoption, many engineering teams remain stuck in fragmented experimentation. This is the opinion of Eric Paulsen, field CTO at Coder.
Paulsen suggests that without a shared understanding of maturity, AI adoption in software development can quickly outpace policy, security and platform controls. He says that this gap makes it difficult for engineering leaders to prioritize investments, demonstrate progress to executives, or safely expand the use of AI agents inside engineering workflows, beyond limited pilots.
“Without a tangible way to understand maturity and governance readiness as a baseline, it becomes difficult to scale agentic AI safely or predictably. Our self-assessment gives teams a concrete view of where they stand, so they can plan adoption intentionally, manage risk and scale with confidence,” Paulsen added.
Coder’s AI Maturity Curve proposes to help organizations frame progress from early experimentation with AI-assisted coding to more advanced, governed use of agents across development workflows. The AI Maturity Self-Assessment builds on this framework by benchmarking maturity across development practices, operational controls and governance readiness.
A Gap Between Intention & Production
Chris Armstrong, DevRel test advocate at SmartBear, has opinions to share in this space. He says that the maturity question isn’t ‘how much AI are we using?’… It’s a question of ‘can we maintain the integrity of our applications as generation accelerates?’
“Teams racing ahead without asking ‘do we understand what we’ve built?’ are creating a gap between intention and production that compounds over time. Mature adoption means shared responsibility: Practitioners asking why, teams normalising questions about intent, leadership resourcing for understanding, not just velocity. Without understanding to bridge it, policy and security can’t close the gap between intent and outcome,” clarified Armstrong.
Roman Zednik, EMEA and also field CTO (it must be a practicality thing) at AI-automated software quality testing company Tricentis, reminds us that as organisations build AI agents that can design, modify and validate software, maturity isn’t defined by autonomy alone.
“AI is only as mature as the data and controls behind it, so without human oversight, even maturity assessments can reflect bias or blind spots. Understanding where agents should act independently and where human validation and testing are essential. This helps teams apply AI responsibly while protecting quality, reliability and trust as AI-driven development scales,” said Zednik.
Beyond Formal Engineering Pathways
Alexander Feick, vice president at eSentire Labs, also has much to say here, and he addresses Coder’s developments directly.
“Coder’s maturity curve asks organisations whether AI adoption is governed, not just widespread. However, it must also ask, ‘Where is that adoption happening?’ The curve assumes AI shows up through formal engineering pathways. However, development is happening outside of them. Agentic tools let non-developers assemble workflow automation that behaves like software and often beyond platform controls, logging and review,” Feick.
Feick also highlights the fact that maturity planning has to account for shadow AI explicitly. This should span from the point of inventory analysis to determine where agents run, understand how we should tier decisions by risk… and also require evidence and traceability to then design escalation paths before scaling. Otherwise, says Feick, we are measuring maturity in places we already control, while the real decisions are being built (and made) somewhere else.
The First Wave of AI Enthusiasm
Dominik Tomicevic is CEO of Memgraph, a company known for its in-memory graph database for real-time streaming data. Tomicevic says that anything that can help organisations improve their level of AI maturity should be welcomed. “This is especially so now, at a time when the first wave of enthusiasm for AI may be hitting a few roadblocks and CIOs might be coming under pressure to justify significant further AI investment. I think any developer team working in practical enterprise IT, especially via GraphRAG, might want to look at this,” said Tomicevic.
Coder’s new offering is a free online tool that maps responses to the maturity curve. The company insists it will help organizations understand their current states, identify gaps and plan next steps for scaling agentic AI. Engineering leaders and platform teams are encouraged to take the assessment and download the maturity curve resource to support internal evaluation, leadership discussions and planning for the next phase of AI-driven software development.
