Hugging Face's logo

Spaces:
sabazo
/
insurance_advisor_wb
Sleeping

App Files Community
insurance_advisor_wb
File size: 849 Bytes 
02c5e7e
 
 
 
 
 
 
 
 
 
2d0bd93
f66560f
2d0bd93
fb95c43
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8761149
f66560f
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
 
---
title: Insurance Advisor Agents PoC
emoji: 🤖
colorFrom: gray
colorTo: gray
app_file: app.py
pinned: false
---


# Insurance Advisor Agent(s)

Setup a modular, multi-agent system to handle inqueries to an insurance company. The system utilizes different approachs to find reliable answers regarding the insurance products

1. Improve Vectorstore search
    Isayah:
    - multi index (chunks+summaries)
    - metadata filtering
    Karan:
    - re-ranking
    - hybrid search (bm25+vectorstore)

    
    - HyDE/HyQE
2. Use ReAct agent instead of chain [DONE]
    - add tools (Google search engine) [DONE]
    - query re-writing
    - ...
3. Use Corrective RAG
4. Workflow for the agents
    - determine funnel stage of customer
    - different prompts per step
    - follow a "script"

This project is licensed under the MIT License: LICENSE.