Google has released the open source framework for creating an AI agents. This framework is called as - Agent development kit.
ADK makes it easy to create, build, deploy and debug the AI agents on the web. It is available in variety of the programming languages such as - typescript, go, python, kotlin and Java. I happen to have learned how to use the Typescript, Go and Python.
Today I want to show what I have built with the help of the Google ADK in Typescript language.
If you are interested in watching the whole journey of launching the agent with the help of Google Antigravity IDE, check the video below.
So I had this problem statement where Solar system needs a chat agent for the students. Like take questions from the students. And the agent has to fetch the content from the Wolfram Alpha. By adding LLM support on the top we can always have up to date answers.
The agent is broken down into the component. We use the Vercel's eve for the chat frontend for the agent. And then the Google ADK is connected to the wolfram alpha if the credits run out we can use the Openrouter API for the queries.
Turns out the agent worked out nicely. I can make the queries specific to the planets in the solar system. And if an off topic queries are made the agent can get us back to the Solar system topic.
I am going to try out another agent with the Python and Golang SDK's. I'd choose another different agent niche next time.
Top comments (0)