AI Blueprint Agent

• 

  This is the architecture diagram of StructifyAI, that visualizes the working of all the agents in automation.

Inspiration

The inspiration behind the AI Blueprint Agent comes from the increasing need of simplifying everyday tasks and assisting the users to complete their works by allowing our automation agent to act as a guide and generate actionable solutions to their queries.

What it does

Our agent communicates with the users to analyze the main goal of the user. It allows the users to enter a link to their audio file, and use the transcriber tool to convert the audio to text format. The agent also extracts key information and entities from the texts using Natural Language Processing. Using this information, an architecture diagram is generated according to templates. Finally, a detailed set of instructions, or in other words, a roadmap is created for the user according to tasks to ensure clarity with their actions.

How we built it

We used the transcriber tool to convert the provided link to the audio into a text format. After the transcription, the text was passed to the NLP function extracts the main entities and the context to understand the need of the user. The logics were used on these parsed data to map and visualize the design in terms of an architecture diagram. Lastly, a set of instructions are generated by the agent, and a brief summary is provided. This agent was orchestrated in the Vertex AI Agent builder, where the instructions handled the tasks and parameters.

Challenges we ran into

Some challenges we ran into in this project are as follows,

• The speech of a person may vary from one another, thus the unpredictable speech frequencies resulted in errors.
• Understanding how to use the tools and functions in the Vertex AI Agent Builder. -Testing and validation of the agent.

Accomplishments that we're proud of

We are proud that we have come to a deeper understanding of how an agent works, and the steps we must follow to create our own agents, as per the input and output needs.

What we learned

-We have learned how to build an agent using Vertex AI.

• We have also realized how a detailed and concise instruction plays a major role in the functioning of an agent. -Integration of pre built tools and function into our multi- agent.

What's next for the AI Blueprint Agent

As we have gained an exposure in building agents, we want to further develop the AI Blueprint Agent by creating an interactive user interface that can be understood by all the people. We also want to take our time to improve the accuracy of the results.

Built With

• dialogflow
• google-cloud
• json
• natural-language-processing
• vertex-ai