Video Agents and Embed Enhancements for the Agentic Web — May 21, 2026

Today, I focused on enhancing the video agent functionality and improving the embeds across the platform, which took an extensive nine hours to complete. The key features shipped included a lead capture popup for video agents, enhanced embed functionalities, and various improvements to the Tavus video integration. These changes are vital as they not only streamline the user experience but also improve the overall effectiveness of our agentic web platform.

One significant feature I implemented was the lead capture popup that appears when a user interacts with the video agents. This addition is designed to capture user information seamlessly while they engage with our content. It’s a crucial step toward building a more interactive and user-focused experience. I spent considerable time ensuring that this feature integrates smoothly with our existing systems, making sure it doesn't disrupt the flow of interaction.

I also tackled some critical fixes to the embed drawer and video connections. The embed drawer now features a full Portal Details panel, which enhances users' ability to navigate and manage their experiences on the platform. I had to diagnose and resolve a few tricky issues, particularly around the iframe permissions for video embeds. The error messages were cryptic, but after some digging, I was able to establish the correct configurations for our X-Embed-Host-Origin settings. This fix was essential in ensuring that our video embeds function reliably, especially when interacting with external sites.

A notable improvement was made to the Tavus video integration, where I added controls for the microphone and a 'Leave' button. These controls are essential for user autonomy during video calls, allowing users to manage their audio settings effectively. It was a challenge to get the microphone permissions working correctly within the iframe context, but after a few iterations, I nailed it down. This feature will undoubtedly enhance the usability of our video calls, making them more user-friendly.

Additionally, I worked on the security aspects of the platform, particularly around the MapLibre integration. I implemented some security hardening measures and improved the user experience during call hangups. These changes are critical as they ensure that our platform remains secure while providing a seamless user experience. I spent a good chunk of time testing these features to ensure they worked as intended without introducing new bugs.

Reflecting on the day, it's clear that building this platform solo with the help of AI tools like Claude Code in VS Code has its ups and downs. While I sometimes miss having a larger team to bounce ideas off of, the autonomy allows me to iterate quickly. The AI assists me in generating code snippets and debugging, but the final implementation and testing still take a considerable amount of time. Today was a testament to that, as I navigated through several bugs and issues that made the day feel longer than anticipated.

The progress I made today is another step toward my goal of building a one-man-show company valued at a billion dollars. Each feature I ship not only enhances the platform but also contributes to the long-term vision of the agentic web. As I continue to push toward this ambitious goal, I find that every small win adds up, and today felt like a significant leap forward in enhancing our video capabilities.

In summary, today was all about refining the user experience with video agents and embeds, addressing bugs, and ensuring that our platform remains robust and user-friendly. As I look ahead, I’m excited about what’s next on the roadmap and how these improvements will drive engagement and satisfaction for our users. Building this platform is a marathon, not a sprint, and each day brings new challenges and opportunities for growth.