Skip to content

Testing multiagent AI flows for CRUD, search, and document construction from databases

Notifications You must be signed in to change notification settings

drewbloom/multiAgentWorkflow

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Multi-Agent Workflow Demo

Welcome to an in-progress demo for a multi-agent setup.

Current Status

  • Streamlit app has authentication and can handle both text and speech modalities
  • AI assistant has tools to understand and query the attached database to help the user

Next steps:

  • Save conversation memory and cache database knowledge for each user
  • Integrate with SharePoint data environment to expand assistant capabilities
  • Create a vector store of the database as an additional tool for quick retrieval
    • This will be necessary for larger dbs, or dbs containing large documents

Overall Goals

  • Offer a user both text and speech modalities to accomplish tasks utilizing their own data
  • Use common Python dependencies to connect together multiple AI agents that work together to determine how best to serve the user
  • Include tool use that allows those agents to engage in semantic search and make SQL queries
  • Integrate with sample data sources so this model can be replicated easily with other databases using SQL

Replicating This Setup

  • Must use a SQLite database or db file
  • app folder contains deployments
  • agents folder contains the agent setups
  • database folder contains the database and scripting for interacting with it
  • This may be all be migrated into utils.py once an appropriate setup is determined

About

Testing multiagent AI flows for CRUD, search, and document construction from databases

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages