Skip to content

maxlhElaia/vc_exec_assistant

Repository files navigation

Data-Driven Hackathon

Midas AI: VC Executive Assistant - Team 8

alt text

This project was built as part of the Data-Driven VC Hackathon organized by Red River West & Bivwak! by BNP Paribas

We belive agentic behavior will take Venture Capital into the AI era. To support VCs in their day-to-day work, we created an agentic framework that provides the basis for agentic behavior in daily operations.

Features:

  • automated monitoring of portfolio companies (or targets) in any funnel
  • fully integrated into the CRM
  • automatic monitoring of the ecosystem
  • agentic creation of signals/alerts with proposed course of action

Sources for Signals:

  • 🧑‍🧑‍🧒‍🧒 headcount monitoring via Harmonic
  • 📂 open positions monitoring based on PredictLeads
  • 📊 reporting analysis based on OpenAI
  • ⚔ competitor monitoring based on similarweb
  • 💬 tracking of mentions around the web

Architecture

This repo is built with layered/onion architecture. The domain module defines all domain objects. On top of this we have services. Services can create signals, i.e. monitor a specific company. They also react on signals and propose course of action (agents). Finally, the signals with the agentic actions get pushed back into the CRM.

architecture diagram

Installation and Usage

Set up the environment file:

ATTIO_ACCESS_TOKEN=
ATTIO_COLLECTION=
OPENAI_API_KEY=
SIMILARWEB_API_KEY=
HARMONIC_API_KEY=
PREDICTLEADS_API_KEY=
PREDICTLEADS_API_TOKEN=

The app is fully dockerized, you follow the steps above and then run:

docker compose build
docker compose up

Alternatively you can create a virtual environment:

python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

To run it locally, create the virtual environment and then run:

python -m app.cli

About

Data-Driven Hackathon

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published