A question we've heard from many clients recently is, "We bought expensive Generative AI licenses for all our developers, but we're not seeing the benefits. What's going wrong?"
It’s a valid concern. The hype around GenAI is massive, but the return on investment can often feel elusive. Here at Devoteam, we've been experimenting extensively with GenAI tools for code modernization, new feature development, and more. Our experience has led us to a clear conclusion: simply giving a developer a license for an AI chatbot is not enough.
The real value of GenAI isn't unlocked by asking it questions in a browser tab; it's unlocked through deep integration into the developer workflow. The intuition arose that using a chatbot for code snippets is just the first step of a much longer journey.
To help developers and their managers understand this journey, I’ve developed a 4-level Developer AI-use maturity model. This model frames a developer's GenAI usage based on two key factors: how integrated AI is with their tools and how widely they use it to create impact.
Let's break down the levels.
Level 1: The Explorer - Conversational Assistant
This is the starting point for most developers, especially if their company doesn't provide official AI licenses. At this stage, GenAI is used as a supercharged search engine or a conversational sounding board.
Developer Mindset: “Instead of Googling or checking Stack Overflow, I’ll ask ChatGPT.”
What this looks like:
Asking for basic code generation for isolated functions.
Pasting a confusing code block and asking for an explanation.
Brainstorming different approaches to a problem.
This level is useful, but it barely scratches the surface. It provides isolated answers without any context of the broader project, and it doesn't fundamentally change how a developer works.
Level 2: The Builder - Active Coding Partner
This is where the magic starts to happen. The developer moves from a separate chat window to an AI tool embedded directly within their code editor (like VS Code or IntelliJ). The AI becomes an active partner in the coding process.
Developer Mindset: “My AI tool helps me write code faster and with fewer bugs. It handles the boilerplate and helps me think through problems in real-time.”
What this looks like:
Getting in-line code suggestions and auto-completions.
Using the AI to automatically fix bugs or syntax errors.
Generating unit tests for a specific function or class.
How to get here: This level is enabled by embedding GenAI in your IDE. However, a major risk emerges here. If your company doesn't sanction specific AI tools, developers might copy-paste proprietary company code into a free tool. This can create serious legal and security issues, as that code could be used for training the public model. A sanctioned, integrated tool is essential to safely reach this level.
Level 3: The Integrator - Whole-Project Awareness
This level is a massive leap forward and simply cannot be achieved with a free browser chatbot. Here, the AI assistant has awareness of your entire project or workspace. It can reason about connections between different files, understand your existing patterns, and help with complex, multi-file tasks.
Developer Mindset: “My AI tool understands my entire project. I can trust it to help me implement complex features and refactor code safely and efficiently across multiple files.”
What this looks like:
Drafting all required files for a new end-to-end feature (e.g., controller, service, repository, and test files).
Performing complex, workspace-wide analysis and refactoring.
Generating technical analysis or documentation for a feature before development even begins.
Creating diagrams that explain the architecture of a large legacy system or a fleet of microservices.
Customization: The AI can be trained on your team's specific coding standards and practices to ensure all generated code is consistent and follows best practices.
How to get here: This requires powerful, integrated tools like Amazon Q Developer. By enabling workspace-wide context (often with a simple @workspace command), the tool indexes your entire codebase to provide deeply contextual answers and actions. As always, you must critically review all AI-generated output, but the time saved navigating a complex project is immense.
Level 4: The Strategist - Lifecycle Optimizer
The final frontier of Developer AI-Use maturity is integrating it beyond the IDE and into the entire software development lifecycle (SDLC). The AI becomes a strategic partner that proactively improves code quality, security, and deployment processes.
Developer Mindset: “Our AI tool is a foundational part of our development lifecycle. It helps us build secure, high-quality, and well-architected applications from the start.”
What this looks like:
CI/CD Pipeline Integration: The AI can help create, debug, and optimize your deployment pipelines. A huge value-add here is that the AI can work on different branches to suggest fixes without the developer having to constantly pull, test, and context-switch.
Proactive Security Scanning: The AI automatically generates merge requests to fix identified vulnerabilities.
Automated Code Quality Reviews: The AI acts as an automated reviewer, checking for best practices and potential bugs before a human even sees the code.
Architecture Decision Support: The AI can analyze proposed changes and provide insights into their potential impact on performance, security, and cost.
How to get here: While this is an area we are still actively exploring, tools like GitLab Duo are leading the charge. They are building AI capabilities across the entire DevSecOps platform, from issue creation to code review and security monitoring.
Where Are You on This Journey?
So, back to the original question: why aren't your AI licenses providing value? The answer is likely that your developers are stuck at Level 1.
The path to unlocking the true potential of GenAI is a conscious journey of deeper integration. It requires not just a license, but the right tools, safe practices, and a mindset shift from using AI as a simple chatbot to embracing it as a strategic partner.
What level is your team at? What tools are you finding most effective? Share your experiences in the comments below!
