Agenda and AI paradigms

0:00 Uh I've uh previously worked at uh Twitter and Amazon. uh and currently I'm a machine learning tech lead at a late stage startup uh driving efforts to apply deep learning, generative AI and agentic AI to personalize conversations in uh uh commer uh commerce space. Um so that's a little bit about me. uh let's move on to uh learn about what is agentic AI. So uh let's take a quick journey through the last decade uh of AI.

0:51 Uh this slide essentially captures a powerful story uh of how machines move from simply recognizing images uh to making decisions and using tools. in just over the last 10 years uh uh 10 plus years or so. So in uh 2012 uh everything kind of changed uh with the advent of alexnet uh which was a deep learning model that drastically improved image recognition and suddenly uh machines could see with near human accuracy. Uh we call this phase the beginning of perception AI uh where systems learn to process and interpret uh visual and textual data.

1:43 Over the next few years, AI was no longer uh limited to just classification tasks. It could now analyze, interpret and even respond to information. uh teams at Google Deep Mind, MIT, CCL, Stanford and others uh kind of pushed the envelope further. The next big leap was within like generative AI in uh late 2022. Chad GPT was brought into this generative AI field. Um and then suddenly AI wasn't just like analyzing content. It was able to like create it.

2:33 Platforms like hugging piece, copilot, enthropic and metaslama help them uh critize these kind of capabilities. Um and now uh 2023 to the present year we are now entering in the agentic AI era. These system don't just like generate content. They can act autonomously, make decisions and use different tools to complete uh various tasks. So think of these uh AI agents uh as uh like a personal assistant which can take actions on your behalf and then that's where in that context we'll kind of learn about tools like uh langraph which can help you build these kind of