The previously mentioned JAA has been operating continuously for six months and has gained a stable user base. Tomorrow, I need to attend the company's OPEN AI DAY to share about JAA in Shanghai.
If it were half a year ago, I would have been very willing and had a lot to say, but now, I think it's already somewhat outdated.
In the previous post, I introduced Step-Copilot, and I stop the development of it at November 2024.
Then, I thought how great it would be if AI could call the tools we had developed?
We had previously built many tools, such as batch operations for Jira, Jenkins maintenance, and others, which provided services to users through GUIs or web interfaces. But regardless of the form, there is always some learning cost involved.
Even with comprehensive documentation provided, most users would likely ignore it and choose to directly ask the tool developers instead.
It would be incredibly helpful to have a AI that not only uses all the tools but also teaches users which scenarios to apply each tool in.
Based on this idea, I developed the JAA
In October 2024, under the wave of AI, my main development work shifted from the automation testing framework to the development of AI products. At this point, I only use AI as a search engine.
Base on my experience with the automation testing framework, the first AI product I want to develop is an AI Code generator, it primarily focuses on the generation of automated test scripts. And I named it Step-Copilot.
Sometimes, I need to run a model locally on my machine. This is useful for debugging and testing purposes.
This guide will show you how to run a model locally using Ollama.