PostQode's AI Odyssey: Part 1 - The Awakening
I wanted to share the AI journey we embarked on with PostQode and how it transformed the platform from a promising low-code API testing tool into something far more powerful.
In early September, we were tirelessly working on the initial version of PostQode. Initially, our plan was to incorporate a few cool AI features into the product later in the development cycle. However, something happened that completely shifted our trajectory. The world of AI began to explode with new possibilities.
At first, we dabbled with new frameworks like AutoGPT and BabyAGI, but it was more out of curiosity than anything else. To be honest, we thought they were more like cool side projects than game-changers.
But then, we stumbled upon some groundbreaking research papers around August to September 2023 that stopped us in our tracks. Papers like AutoGen, ChatDev, Gorilla and RestGPT painted a picture of AI that was far more powerful and versatile than we had ever imagined. It was a lightbulb moment – we realized that AI could do more than just predicting text; it could actually enhance the entire way developers interact with our platform by completing tasks and improving overall productivity.
Our first step was to immerse ourselves in the research papers and explore various agent frameworks and large language models (LLMs). We conducted multiple Proof of Concepts (PoCs) using frameworks like LangChain and LlamaIndex, along with other libraries and opensource models that were gaining traction. We spent considerable time understanding their capabilities and limitations.
This newfound understanding made us reevaluate our strategy for PostQode. By September-October, we were confident that AI could significantly enhance our platform and industry. We recognized an opportunity to leverage AI to assist developers seamlessly, and we didn’t want to miss it. Despite being a bootstrap company, we made the bold decision to pause our planned releases and invest more in AI integration.