Application Development

June 05 , 2024 by Sean Anderson

Overview

We hear a lot of buzz about generative AI and LLMs but see fewer actual projects reaching beyond a simple prototype. Generative AI is positioned to create over $200B in business opportunities for companies that leverage it correctly, and the use cases abound. We launched an initiative leveraging the giant brains in our community to showcase their best project that uses Vectara GenAI. We were overwhelmed by the response and interest, and even in the short month period, we received eight extremely impressive projects. Still, three really stood out in their ability to showcase the power of generative AI.

The contestants were able to harness the power of the Vectara platform without restrictions as we opened up all the popular Scale features to participants in the contest. This meant they had access to the full suite of tools inside the Vectara platform, such as access to the powerful GPT-4-Turbo for generative summarization, and the ability to use custom prompts. We received submissions solving use cases like integrating with smart glasses, public safety, various calculators, and even ways to engage better with your social media posts. We feel that the three winners really tackled some of the most pressing use cases and provided an end product with functional utility.

The Entries

Wisdom Hoard by Giridhar Talla, Sabih Sarowar, Abdul Mustafa

Introducing Wisdom Hoard. This is a Discord bot you can send links to for interesting social media posts you find online. The bot will save it and allow you to ask it questions whenever you need to, and the bot will answer from your repository of saved posts. 

For example, you can ask, “What makes a good programmer?” and it will give advice based on your posts. You can plan on having this replace your time-consuming way of saving posts, which is often taking a screenshot, messaging the link to yourself via text, or saving it in a folder.

“We used Vectara for it, and it was relatively straightforward, with only two endpoints involved. One for uploading the content of a post and one for passing in a question and retrieving the LLM output.” – Giridhar Talla

About the Authors:

Giridhar is a full-stack JavaScript developer. JavaScript and Python are his favorite programming languages because they allow him to bring my ideas to life in the software development and machine learning domains. 

Sabih is addicted to building with my keyboard clicks, A soul with too many ideas, and the lack of time makes him sick. But every idea, he will make and refine, bettering the world, one line at a time.

Abdul is a Software Engineer with 2+ years of experience across web, cloud, backend, and frontend development.

Robert F. Kennedy Chatbot by Erica Trump

You likely have your own personal thoughts of the enigmatic character who is running for president of the United States as a third-party candidate. Regardless of your personal views, it’s clear RFK has lived a storied existence that manifests into a method of speech that is as branded as the figure himself.

The RFK Chatbot started with content from RFK Jr.’s campaign website and YouTube videos of his interviews, The chatbot was created that espouse his viewpoints. Vectara is central to this project, streamlining the entire RAG process. It automatically handles indexing by splitting source text, creating embeddings, and storing them in a vector database. The platform also manages retrieval, orchestrates generation by integrating with LLMs, and maintains conversational memory—critical components for a chatbot that I previously had to manage step-by-step.

“Vectara’s ability to streamline the chatbot creation process has truly impressed me. For this competition, I was able to quickly and efficiently produce the RFK Jr. Bot using Vectara and its ecosystem tools. Vectara has significantly simplified the complexities of RAG, making it more accessible to people who are applying LLMs and prototyping products, but who aren’t necessarily developers, like myself. By sharing this on LinkedIn, my aim is to showcase Vectara’s capabilities to my network and beyond—especially to those who are working on chatbot products and who are unaware of the value that a generative AI platform can bring.” – Erica Trump

About the Author:

Erica is a technical leader with nearly ten years of experience in the industrial software industry, including seven years of experience in program and product management. I have broad experience in setting program strategy, collaborating with stakeholders across a broad organization, planning technical roadmaps, managing customer success objectives, and team leadership. I have focused on software solutions that support industrial IoT and data science applications for energy, sustainability, oil and gas, food and beverage, and pharmaceuticals. I am a hands-on generative AI expert well-versed in creating RAG pipelines, agents, and chatbots to solve real-world problems.

FirstAid RAG by Emmanuel Owezie

Introducing FirstAid RAG virtual assistant, the AI-powered chatbot that just might save your life. Like many of us, you likely rush to google when you step on a rusty nail or have a wound that bleeds a little longer than you would like. FirstAidRAG gives you the facts minus the hallucinations. It’s a simple chat widget embedded in a health web app. FirstAid RAG provides instant answers in emergency situations to help with first aid treatments. It provides you with the answers and citations so you can read more and feel confident the advice you are receiving is based on real medical research. 

“I had a smooth experience using Vectara. From feeding the corpus with relevant data, to using the prebuilt UIs. Vectara’s developer-friendly platform really makes it easy to embed powerful neural search and chat functionalities into applications.” – Emmanuel Owezie

Conclusion

We were absolutely blown away by what you all created. It truly showcases the power and diversity that generative AI can bring to every user and business if they have a clear picture of their use case. You saw the common threads about ease of use and how quickly it takes to build a RAG pipeline, even without a deep bench of machine learning experience. A hearty congrats to all our contestants. We will be doing more “Built by Vectara” contests so please stay tuned to our Discord Server. Until next time!