A Brief look at how the 'NoteRep AI Chatbot' works

ShravanShravan
5 min read

Hey there! I’m here to share how I integrated a lightning-fast AI chatbot for NoteRep - a platform dedicated to organized study material, notes, PYQs, and much more. Let me explain the technology that powers this tool and explore the impact it has.

The NoteRep Vision: More Than Just Notes

Before I explain the working of Chatbot, let me set the context. NoteRep isn't just an average study materials platform. Sure, it's got notes and resources, but I've also add bunch of features like a calendar for tracking events and exams, community links to all the clubs in our college, a real-time class timetable, shared links for spreading knowledge, and even a countdown feature for upcoming exam. (received mixed feedback for this feature 😅).

But I wanted to take things to the next level. Introducing the AI chatbot – the cherry on top of this feature-packed website.

The Need for Speed (and Smarts)

When I decided to add an AI chatbot to NoteRep, I had two non-negotiable requirements:

  1. It had to be fast. Like, blink-and-you'll-miss-it fast.

  2. It needed to be smart enough to help the students within the set boundaries.

Enter Groq: The Speed Demon

After exploring various endpoints that enable seamless integration with popular open-source Large Language Models (LLMs) from providers like Anyscale, Together AI, and Clarifai, as well as private LLM companies such as Anthropic (Claude) and Open AI (GPT), then I stumbled upon Groq – an AI infrastructure company that promised to deliver mind-blowing inference speeds. And, did they deliver! We're talking about 1000 tokens per second. To put that in perspective, it's like you get the response in a blink of an eye. You might wonder how is it fast so let me explain, so Grok rolled out this thing called an LPU - short for Language Processing Unit. It's a new chip that's all about making AI smarter and faster, especially for chatting or writing like a human. While your computer's brain (CPU) and muscle (GPU) are great at lots of tasks, the LPU is like a specialist who's really good at one job: making AI run smoothly and quickly.

The result? The chatbot can spit out responses in about 0.15 seconds. That's faster than you think.

Behind the Scenes: Smarts with Firebase Firestore

But speed isn’t the only trick up the chatbot’s sleeve. The chatbot doesn’t just "talk" – it logs every interaction using a backend I’ve connected to Firebase Firestore. Here’s how it works:

When a user opens the chat for the first time, a JavaScript function generates a pseudo-random, 18-character string, which serves as a unique user ID. From that point on, every time the user interacts with the AI, their chat history is stored locally in their browser and synced with Firebase Firestore in real time. This allows the chatbot to remember past conversations, helping users pick up right where they left off.

Admin Powers: Monitoring and Control

But there’s more. I’ve built an admin panel that lets me monitor user activity in real time. This isn’t just for oversight; it’s designed to ensure that the chatbot can evolve and improve based on actual usage. For example, I can inject specific system prompts or user prompts into a conversation, guiding the chatbot to respond in a particular way. This level of control ensures that responses remain within the platform’s boundaries while still being highly relevant and helpful to students.

This monitoring feature is especially useful for refining how the chatbot handles ambiguous questions or integrates with other NoteRep features like the calendar, class timetables, or resource sharing. Essentially, I can nudge the chatbot to be more aligned with student needs based on live interactions.

Seamless Ad Integration: Smarter, Subtle Monetization

One of the latest features I’m particularly proud of is seamless ad integration within the chat conversations themselves. Rather than interrupting users with annoying pop-ups or banners, the chatbot carefully slides context-appropriate ads into the conversation. For example, if a user mentions they're bored, the chatbot might recommend nearby entertainment venues, such as bowling or VR gaming spots, driving traffic to those businesses while keeping the user engaged.

Take this real example:

On Sep 17 at 10:01 AM, a user sent a query saying, "I'm bored." The chatbot responded by recommending "The Grid" – a nearby entertainment hub with activities like bowling, laser tag, and VR games, located near Ramaiah College.

By suggesting relevant local services in a seamless, conversational way, we create opportunities for businesses to promote their offerings without disrupting the user experience. Similarly, when users mention study-related stress, the chatbot might recommend nearby cafes or food outlets for a refreshing break—encouraging real-world engagement while subtly integrating monetization.

Final Thoughts: A Powerful Combo of Speed, Smarts, and Strategy

With NoteRep’s AI chatbot, I wanted to push the boundaries of what a learning platform could offer. By combining blazing-fast Groq AI inference, smart interaction logging via Firebase Firestore, and admin-level control over the chatbot’s behavior, I’ve created a tool that’s not just responsive but also highly adaptive to the needs of students. The seamless ad integration ensures that the platform remains free while delivering value to users.

And this is just the beginning – I’m continuously refining and expanding its capabilities. Like I will further want to integrate RAG on all the study materials.

Final Thoughts: Just the Beginning

In a nutshell, the AI chatbot on NoteRep is fast, smart, and designed to make your life easier without getting in the way. Whether it's answering your questions, keeping you on top of your schedule, or subtly pointing you towards something fun, it’s here to help. And with more features on the way, like RAG integration, I’m just getting started. Stay tuned—things are about to get even more interesting!

0
Subscribe to my newsletter

Read articles from Shravan directly inside your inbox. Subscribe to the newsletter, and don't miss out.

Written by

Shravan
Shravan

As a junior student at M S Ramaiah Institute of Technology, I am pursuing a degree in CSE (Spl. AI & ML). My passion for coding started a few years ago, and I have been constantly striving to expand my knowledge and skills, working with a variety of technologies. Although you can find a few of my projects online, some are private too 😉. My goal is to make a meaningful impact by contributing to open-source projects and creating practical solutions that can save time for others. In addition to my technical pursuits, I share my thoughts and knowledge on technology through my blog and enjoy helping others with common tech challenges.