Qodo Gen 1.0: Evolving AI Test Generation to Agentic Workflows
When Qodo was founded in 2022, our philosophy was that quality software isn’t just a product of well-written code, it’s the outcome of solid, rigorous testing and review. Our first AI tool, what’s now Qodo Gen, focused on exactly that—making writing tests, one of the least-liked tasks of software development, easier with Generative AI.
Our vision was to make writing tests feel like magic, effortless and achievable in just one click. We built an IDE solution that automates test creation, and expanded capabilities for code behavior analysis, test coverage, and more.
Fast forward to today, and while our philosophy remains the same, AI has evolved. Developers expect AI to work with them, adapting to their workflows rather than just automating tasks and generating outputs.
With that in mind, we’re excited to introduce Qodo Gen 1.0, featuring a semi-agentic test generation experience. Built for deeper context awareness and seamless integration into developer workflows, Qodo Gen 1.0 provides guidance, adaptability, and control, helping developers write better tests with AI that truly understands context, intent, and best practices.
Qodo Gen 1.0: From automation to agentic
In Qodo Gen 1.0, users now have the option to use a chat-based, semi-agentic test generation workflow. The test agent takes users through a step-by-step flow, giving you options to add example tests as a context, choose mocks and frameworks, and generate test behaviors.
This new agentic experience for test generation is a departure from our original workflow, which features a dedicated side panel within the editor as the central hub for testing. There, users choose a framework, add example tests, and generate code behaviors and tests.
Each test is displayed in a dedicated code block within the panel and an embedded chat interface allows developers to interactively request modifications, refine assertions, or adjust test parameters.
Over time, we expanded the panel’s functionality, adding key capabilities like the ability to execute tests, analyze results, and fix failures.
When we launched Qodo Gen, developers appreciated the interface of this testing panel, along with the speed and accuracy of test generation. They found value in how Qodo Gen analyzed code, identified different behaviors, and learned from their examples to generate relevant tests. Many also noted its ability to uncover edge cases they hadn’t considered and detect potential bugs, helping them refine their code more efficiently.
So why on earth did we decide to revamp our test generation workflow?
Key Advancements in Qodo Gen 1.0’s Agentic Workflow
Three years is a long time in AI—agentic workflows emerged, LLMs became significantly more powerful, and IDE extensions gained deeper access to system resources, enabling more seamless integrations. And we expect AI to do a lot more now than ever before.
Our “magic” one-click test generation worked well for simple projects, handling more complex codebases requires a more adaptive, context-aware approach—one that could leverage deeper integration with the editor and project structure to deliver more precise, useful results.
Our new agentic test is built on few key principles:
- Step-by-step guidance: A structured, intuitive flow ensures developers get one solid test to start with, making it easier to generate additional tests as needed.
- Solid first shot test: when there’s one good test, the rest of the tests and uses cases are easy to generate
- Editor-native experience: Tests are generated inside the IDE, giving full control over everything in the file—imports, helper functions, and dependencies.
- Seamless execution: Tests can still be run directly within the IDE
- Handle mocks more deeply: The workflow includes a step that suggests relevant mocks and allows users to choose a preferred mocking framework, resulting in a better first test.
- Flexible and scalable: The UI and backend were designed for continuous evolution. As the agent improves, the interface adapts. The system is built to support more test types—end-to-end, integration, and UI testing—with minimal effort.
- MCP integration for richer context: Our agent uses Anthropic’s Model-Context-Protocol (MCP), enabling broader contextual awareness. For example, integrating with Postgres MCP could fetch relevant database context, allowing the generation of more accurate DB mocks.
Looking Ahead
As we embrace this new semi-agentic approach to testing with Qodo Gen 1.0, we’re just scratching the surface of what’s possible. Our journey from automated test generation to an intelligent, context-aware test agent represents more than a product evolution—it reflects our ongoing commitment to solving developers’ most challenging testing problems.
The philosophy that guided us in 2022 remains unchanged: quality software demands thorough testing. But our methods have matured alongside advancements in AI. By creating a workflow that adapts to developers rather than forcing developers to adapt to tools, we’re making test creation not just easier, but more intuitive and effective.
We’ve built Qodo Gen 1.0 with flexibility and scalability in mind, laying the groundwork for supporting more complex test types—from integration to end-to-end testing—in future releases. Our integration with technologies like MCP opens doors to even richer context awareness, allowing our agent to understand your codebase at unprecedented depths.
We invite you to try Qodo Gen 1.0 today and experience firsthand how this semi-agentic approach transforms test creation from a mundane task into a collaborative process. Share your feedback with us as we continue refining and expanding this technology—because ultimately, our vision of making testing feel like magic is realized through your success.
Testing shouldn’t be an afterthought or a burden. With Qodo Gen 1.0, it becomes an integral, accessible part of writing exceptional code—exactly as it should be.
