Social Enterprise Business Coach AI

The AI Agent acts as a social enterprise business coach. Using information from Social Sector Network's resources, the AI Agent helps users with business ideation and strategy.

Comment

Inspiration

I wanted to create an AI Business Coach to assist social entrepreneurs who are building businesses aligned with the UN Sustainable Development Goals. Ideally, this AI Agent will enable me to offer a scalable coaching service that is not dependent on my labor.

What it does

Social Enterprise Business Coach AI helps current and aspiring social entrepreneurs with business ideation and strategy for their startups focused on the UN Sustainable Development Goals.

How we built it

I followed these steps to create Social Enterprise Business Coach AI:

  1. Created an AI Agent with Vertex AI
  2. Added a Name, Goal, and Instructions for the Agent
  3. Created a Tool
  4. Created a Website URL Data Store and a Cloud Storage Data Store
  5. Integrated the Tool with the Data Stores
  6. Updated Instructions and Examples for the AI Agent
  7. Added a Dialogflow Messenger Integration
  8. Added the Dialogflow Messenger Code to SocialSectorNetwork.com

Challenges we ran into

  1. Data Stores - I initially created the data stores outside of the app for the agent, so the agent was unable to query the data stores during testing. I was able to resolve this issue by clicking the link to create data stores in the tool interface then making data stores associated with the correct app. I then had to refresh the tool page and associate the tool with the data stores.
  2. Dialogflow - When I initially set up the Dialogflow integration, I did not get any responses in testing. I was able to resolve this by reading through the Dialogflow Messenger documentation and using an Unathenticated API.
  3. Agent Responses - I discovered a problem with the agent where it would query the data store and find resources to answer the user's question. It would then tell the user "Here is a list of resources" but not provide the resources in the user-facing response. A workaround on the user side was to tell the agent "You did not provide the resources" so that it corrected the mistake. Additionally, I decreased occurrences of this issue by editing the agent response in the examples.

Accomplishments that we're proud of

I am excited that I was able to make an AI Agent and integrate it with a website.

What we learned

I learned how to create an AI Agent, set up data stores, refine agent responses, and implement a website chatbot.

What's next for Social Enterprise Business Coach AI

I plan to gather user feedback, add more data stores, and improve the AI Agent.

If you would like to try the chatbot, you can go to SocialSectorNetwork.com (see link below) and use the coupon code "VertexHack64" at checkout. The code will provide free access through June 25, 2024.

I hope you find the chatbot useful in your business endeavors!

Built With

Share this project:

Updates