Generative AI Hackathon with IBM watsonx

Start

Providing a virtual VC Pitching Simulation utilzing IBM Granite LLM with RAG (with SingleStore)

VC Simulator

Abdo Magdy

Angel Invesors, VCs, Startup Accelerators and Scouts Sift though large quantities of contexual data provided by startup seeking investment a very efficient process that can be optimized with GenAI.

This solution is based on IBM's Prompt Lab structured input and output. The model is prompt with chuncks of texts from the provided pitch deck and instructed to extract specific  data points based on given examples, the extracted data points are then integrated in the solution with the help of python to calculate the most important indicators for startup state such as Burn Rate, Customer Lifetime Value, Customer Acquisition Cost, Market Size.

The model iterates across a given "startup profile" format to ensure completion of data and to prompte the user for any missing details before creating a dashboard showcasing the calculated variables with reference to the source extracted.

In an advanced version, RAG implementation with Semantic Search could make this solution more powerful.

This business productivity solution saves a lot of time and effort for a niche market that has to sift through raw startup data.

Furthermore, the solution could be developed into providing the same "investor dashboard" generated to the startup that uploaded it, allowing an interactive pitch simulation session and utilizing a RAG of embedded "startup knowledge base" to exact the startup pitch.

VCs and Startup Accelerators receive tons of startup applications to sift through, this facility uses IBM GRANITE 13B LLM model, to extract starup data points from uploaded pitch deck files and present the most critical indicators in a dashboard 

"VC Simulator team on Generative AI Hackathon with IBM watsonx Hackathon"

Team Idea

Submission