Forging the Path Towards Interactive AI

Yes, I'm coining this as the new hype.

We've seen soooo many "GPT-wrappers" in the last few months and alot of people criticizing ChatPDF because now OpenAI has document analysis builtin so we should all be hesistant to build products on "top of another platform".

Here's the thing - the people preaching this all over LinkedIn, IndieHackers, Reddit, and Hacker News are not seeing that something like ChatPDF is more like a feature that ChatGPT was lacking rather than a full-blown product. And for that, it definitely served its purpose well.

It's very easy to build something like this but it's something that was obviously going to be built with the entire multimodal push OpenAI has been promoting. Primitives and infrastructure is usually built first, which is kinda like how AWS is where it is today.

So actually, I think it is totally ok building something on top of another product if you are focused on building product value rather than just functionality.

Let's face it, alot of startups have landing pages that thrive on saying they have X and Y feature and their competitors only have X feature so clearly, they must be the better product.

It's not the case.

The better company is the one that brings immense value to the users it is intended for. Chamoy is 1000% focused on doing this, despite some of our competitors having more features than us.

We bring this value by focusing on Interactive AI.

Interactive AI is 2024's Buzzword

Interactive AI, often characterized by chatbots, virtual assistants, and interactive recommendation systems, is designed to engage with users directly. It's built to understand user input, provide real-time responses, and offer personalized experiences. On the other hand, Generative AI focuses on creating content, such as text, images, or even music, based on patterns and data it has learned.

We've seen interactivity in certain products such as Siri, Alexa, and Cortana but they were intended to be used as general assistants and that's why they've inherently failed as transforming into popular products. I personally have never used my Google Home other than asking it to play music.

Chamoy's internal AI harnesses the capabilities of GPT-4, and we're also exploring Claude, to create highly immersive experiences. Our aim is to elevate our AI, Ellie, to function convincingly as a recruiter and interviewer. Advanced prompt engineering techniques amplify the authenticity of these interactions significantly.

Prompt engineering techniques involve the careful design and construction of input prompts or queries provided to the AI model. The idea is to tailor these prompts in such a way that they produce desired and realistic responses.

This process can include fine-tuning the prompts, experimenting with different phrasings, and considering the context in which the AI will be used. It's about finding the right balance of specificity and generality in the prompts to generate coherent and contextually relevant responses.

Prompt engineering is a critical aspect of refining AI interactions, as it directly impacts the quality and authenticity of the output. It's an ongoing process, and by continuously iterating and optimizing prompts, the AI system can become more effective and convincing in its role as a recruiter and interviewer.

Why Personalization Actually Matters

Interactive AI is designed for real-time engagement, making it an ideal choice when direct interaction with users is a priority. The instant nature of these interactions is invaluable in scenarios where quick responses are essential such as interviews. We parse, extract, and summarize contents of the candidate's resume prior to generating questions so that its completely tailored to the candidate.

Alot of interviewing platforms today don't have this fine-level of personalization so the interactive experience is diminished on those platforms. In fact, most of our competitors simply ask the user what industry they are in or what role they are seeking and generate based off that. Unfortunately, that still yields for cookie-cutter generated questions because it doesn't take into the candidate's experiences into account.

Two Java programmers that worked in different industries within different companies with different years of experience should not be presented with the same set of questions at all.

We're also working on personalizing the recruiter as well with using more realistic text-to-speech for communication, actively seeking to integrate Bark, and offering the candidate customizations to the recruiter based on company values and behavior.

Stop Shitting All Over The Internet

If you use any form of generative AI to generate info-based content like blog posts, writings or copy, then you're actually contributing to polluting the internet. Think of this as littering. Flooding the internet with an excessive number of low-quality, generated AI posts can be considered a form of digital pollution. Clogging the online space with irrelevant or unhelpful content but also strain the resources of platforms and potentially hinder access to valuable information for literally everyone.

In interactive AI applications, it is crucial to uphold a high standard of content quality and relevance to ensure a positive user experience. So developers in this space (like myself) are continually pushing experimentation to create personalities that are engaging and fun to work with.

This is how you can maintain high levels of adoption without much fear of churn. People will always want to work with those applications that are easier to work with and have excellent UI/UX. Accessibility and usability directly impacts user satisfaction because good UI/UX helps users understand that the product was truly built for them.

A Single Closing Thought

There's no competitive edge anymore by integrating AI into your product. It doesn't matter if you're an established company or fledgling startup. Everyone has AI now. Focus on forging interactivity first.