Virtual Assistants 2.0: How Gen AI elevates the game of domain-specific virtual assistants

Amar Jaiswal (~amar3)


45

Votes

Description:

Problem

Traditional virtual assistants like google's DialogFlow provide fixed responses, because of which the conversations with then is really bot-like. They cannot simplify or summarize or translate their responses, because of which the UX very monotonous. Generative AI solves this exact problem and makes the responses very human like and skyrocketing the UX.

Summary

Domain-specific virtual assistants using Gen AI! πŸ€–πŸ’‘ πŸ“£ Discover how we've harnessed the power of Vector DBs and Embeddings to overcome context limitations and create cost-optimal solutions for enterprise products.πŸ”πŸ’° πŸ€” Traditional chatbots fall short when it comes to generative capabilities as they only give fixed responses. But fear not! πŸ™…β€β™‚οΈ Our talk explores the challenges we faced while fine-tuning and completing models, and how we revolutionized the game. πŸš€ ⚑️ Don't miss out on the discussion of our powerful combination of embeddings, vector DBs, and chat completion models, which not only inherit the strengths of generative capabilities but also solve the dreaded context/token limitation! πŸ’ͺπŸ”“ 🌐 Join us for an engaging talk where you'll learn how to empower your customers and teams, and revolutionize the world of virtual assistants! πŸš€πŸ’Ό πŸ—“οΈ Save the date and be prepared for an enlightening session that will reshape the way you approach AI! πŸ“…βœ¨

About the talk

  • This talk is about Using Gen AI to build domain specific virtual assistants for enterprise products.
  • Little to no prior AI/ML experience required to built the solution.
  • What all challenges we faced while building a usable solution.
  • How Vector DBs and Embeddings helped us resolve the problem of context limits and made the solution cost optimal.

Outline / Topics covered

  • Why traditional chatbots are not enough?
  • How Gen AI solves the problems / drawbacks of traditional chatbots
  • Challenges faced by us while building the solution
  • Our Solution / High Level Architecture / Framework
  • Working Demo
  • Similar Alternatives available in the market.

Expected Outcome

Folks looking to build next gen virtual assistants or enhancing the capabilities of their product will have a clear understanding of how they can use GenAI to achieve their goal.

Prerequisites:

Basic knowledge about Chat GPT and LLMs required. Basic knowledge of what vector databases is required.

Speaker Info:

Hey there! πŸ‘‹ I'm Amar Jaiswal, I am a Senior Backend Engineer at Certa and leading AI Initiatives in the product leveraging the power of Gen AI. From coding to tinkering with gadgets, I love diving into the world of technology and exploring its endless possibilities. I'm a tech enthusiast with a knack for making things happen and I like to play video games.

Section: Data Science, AI & ML
Type: Talks
Target Audience: Intermediate
Last Updated: