<?xml version="1.0" encoding="UTF-8"?><oembed><type>video</type><version>1.0</version><html>&lt;iframe src=&quot;https://www.loom.com/embed/536cb33b3d7d4af5a32022c5fa0c1604&quot; frameborder=&quot;0&quot; width=&quot;1668&quot; height=&quot;1251&quot; webkitallowfullscreen mozallowfullscreen allowfullscreen&gt;&lt;/iframe&gt;</html><height>1251</height><width>1668</width><provider_name>Loom</provider_name><provider_url>https://www.loom.com</provider_url><thumbnail_height>1251</thumbnail_height><thumbnail_width>1668</thumbnail_width><thumbnail_url>https://cdn.loom.com/sessions/thumbnails/536cb33b3d7d4af5a32022c5fa0c1604-095c73c3617097c4.gif</thumbnail_url><duration>237.259</duration><title>Reinforcement Learning and On-Chain Data: A New Framework for AI Development</title><description>In this video, I introduce FlowRL, a framework that leverages reinforcement learning (RL) to enhance large language models using on-chain data. I believe that as commerce increasingly moves on-chain, having a general-purpose RL framework will allow users and protocols to monetize their data effectively. I demonstrate this by showcasing a game called Word Hunt, which I deployed on Flow&apos;s testnet, where an AI agent interacts with players to optimize their scores. I encourage viewers to consider the potential of integrating such frameworks into their blockchain projects, as it could lead to significant advancements in AI and user engagement. Let&apos;s explore how we can harness this technology together!</description></oembed>