Kai (/kaɪ/, rhymes with pie) - An AI-enabled tool that simplifies the process of modernizing application source code to a new platform. It uses Large Language Models (LLMs) guided by static code analysis, along with data from Konveyor. This data provides insights into how the organization solved similar problems in the past, helping streamline and automate the code modernization process.
Kai is an AI-enabled tool that assists with modernizing applications. Kai is designed to help developers write code more efficiently by providing suggestions and solutions to common problems. It does this by performing Retrieval Augmented Generation (RAG), working with LLMs by using Konveyor analysis reports about the codebase and generating solutions based on previously solved examples.
Now, you may be thinking: How is Kai different than other generative AI tools?
Konveyor generates analysis reports via Kantra throughout a migration. This history of reports tells you what’s wrong with your codebase, where the issues are, and when they happened. This functionality exists today, and developers are already using this data to make decisions. And because of our RAG approach, this is all possible without additional fine-tuning.
As you migrate more pieces of your codebase with Kai, it can learn from the data available, and get better recommendations for the next application, and the next, and so on. This shapes the code suggestions to be similar to how your organization has solved problems in the past.
LLMs are very powerful tools, but without explicit guidance, they can generate a lot of garbage. Using Konveyor’s analysis reports allows us to focus Kai’s generative power on the specific problems that need to be solved. This pointed, specific data is the key to unlocking the full potential of large language models.
Note
Kai is in early development. We are actively working on improving the tool and adding new features. If you are interested in contributing to the project, please see our Contributor Guide.
- See ROADMAP.md to learn about the project's goals and milestones
- See docs/evaluation_builds.md for information on early builds.
- Technical background for our approach
- Initial presentation slides introducing Kai
- See other technical design related information at docs/design
- 2024 August 29: Incident Storage in Kai - A Deep Dive
- 2024 July 23: Embracing the Future of Application Modernization with KAI
- 2024 May 07: Apply generative AI to app modernization with Konveyor AI
- 2024 May 07: Kai - Generative AI Applied to Application Modernization
Check out our 15 minute guided demo video to see Kai in action!
There are two elements to Kai that is necessary for it to function: the backend and the IDE extension. The backend is responsible for connecting to your LLM service, ingesting static analysis reports, and generating solutions. The IDE extension is where you can interact with Kai, see suggestions, and apply them to your codebase.
- Git
- A container engine such as podman or docker. We will provide instructions for podman.
- Docker-compose or Podman-compose, either one will work. For podman-compose, you can install it here.
Important
Kai is in early development and is not yet ready for production use. We
currently recommend checking out the git tag stable for the most stable user
experience.
The quickest way to get running is to leverage sample data committed into the
Kai repo along with the podman compose up workflow
git clone https://github.com/konveyor/kai.gitcd kaigit checkout stable- Make sure the podman runtime is running with
systemctl --user start podman - Make sure the
logsdirectory accessible to the podman container withpodman unshare chown -R 1001:0 logs- This is necessary to allow podman to write to the
logsdirectory outside the container. - Use
sudo chown -R <your_user>:<your_user> logsto change the ownership of thelogsdirectory back to your user when done.
- This is necessary to allow podman to write to the
- Run
podman compose up.- The first time this is run it will take several minutes to download images and to populate sample data.
- After the first run the DB will be populated and subsequent starts will be much faster, as long as the kai_kai_db_data volume is not deleted.
- To clean up all resources run
podman compose down && podman volume rm kai_kai_db_data. - This will run Kai in demo mode, which will use cached LLM responses, via
setting the environment variable
KAI__DEMO_MODE=true. To run without demo mode executeKAI__DEMO_MODE=false podman compose up. See docs/contrib/configuration.md for more information on demo mode.
The Kai backend is now running and ready to serve requests!
After you have the kai backend running via podman compose up you can run
through a guided scenario we have to show Kai in action at
docs/scenarios/demo.md. This document walks through a
guided scenario of using Kai to complete a migration of a Java EE app to
Quarkus.
The above information is a quick path to enable running Kai quickly to see how it works. If you'd like to take a deeper dive into running Kai against data in Konveyor or your own custom data, please see docs/getting_started.md
- Kai backend will write logging information to the
logsdirectory. You can adjust the level via the environment variables. For example:KAI__FILE_LOG_LEVEL="debug". - Tracing information is written to disk to aid deeper explorations of Prompts and LLM Results. See docs/contrib/tracing.md
Our project welcomes contributions from any member of our community. To get started contributing, please see our Contributor Guide.
Refer to Konveyor's Code of Conduct here.
This project is licensed under the Apache License 2.0 - see the LICENSE file for details.