Ai Lead Generator

• 

  Entity extraction after recording the voice/conversation.

Inspiration

I recently went to the Nvidia AI summit and saw so many startups there. They were presenting their ideas and business models to get more leads and customers. In the past, I have worked on audio/video-based translator and summarizer. I thought that the same idea if applied here would help generate more leads.

What it does

When an interested person visits your stall and asks about your startup, you will be recording the entire conversation and transcribing it to text where you will ask the person their details and you won't be writing but just recording it. Then the LLM agent will perform an entity extraction on the transcribed text and fill the form with the details of the lead like name, company, objective, email, etc. It now supports image-based entity extraction.

How I built it

step1: Record the conversation step2: Transcribing to text step3: Entity extraction using LLM step4: Displaying the final generated lead for validation

Challenges I ran into

Just putting everything together to make it work, that was just ughh... well ffmpeg was an issue

Accomplishments that I am proud of

That it worked, even though it's basic I know the potential it has it used properly.

What I learned

The path of an idea to execution is easy only in theory but in practice, we have to find a good balance of both.

What's next for Ai Lead Generator

I can see a bunch of use cases for this agent like in a courtroom for recording cases, making to-do list, shopping lists just by recording, etc. I am thinking of many things to develop it further. There's one critical use case for this: voice activation of software via function calling support. It's a bit advanced, but it is possible with ample time and could be perfect also.

Built With

• llama
• llm
• python
• wishper