DecodeFM

MCP (Model Context Protocol) and A2A (Agent2Agent Protocol) are getting plenty of attention lately. But what exactly are these protocols, and what problems are they trying to solve? In this episode, we'll decode what MCP and A2A have in common, explore the use cases that could benefit from them, and highlight what sets them apart.

Video production has traditionally involved many tedious tasks - things like motion capture, rotoscoping, and precise camera tracking. These jobs typically require significant time, specialized teams, and complex setups. But now, AI-powered tools like Wonder Studio, Runway Act-One, and Runway V3 Alpha are making these tasks simpler, faster, and more accessible. In this episode, we’ll look at how these tools are already reshaping video production, and discuss what the future might hold as AI becomes an everyday part of the filmmaker's toolkit.

In this episode, we'll explore the journey of character.ai. Founded in 2021 by ex-Googlers Noam Shazeer and Daniel de Freitas, character.ai quickly grew into a popular AI chatbot platform, attracting tens of millions of active users and raising hundreds of millions in funding. Then, in August 2024, it struck a strategic deal with Google - almost an acquisition. What exactly happened? Let’s find out.

AI agents are a hot topic right now. We've talked about different aspects of agents before - like the operator, browser use, and deep research. In this episode, we're looking at a new agent that's getting a lot of attention: the "Super Agent" from Genspark. Genspark claims they've reached $10 million ARR in just 9 days. By my calculation, that's about $250K in revenue for those 9 days - not exactly "annual" recurring revenue (yet), but still impressive. So, what's special about Super Agent? What can it actually do, and how does it compare with other AI agents out there? Let's get...

CoreWeave is a cloud computing company focused on GPU infrastructure tailored for AI workloads, and it just went public recently. In this episode, we'll cover CoreWeave's history, discuss its competitive strengths and potential weaknesses, and compare it to traditional cloud providers. Can it stay ahead of the competition? Let's find out.

A hot new open-source project, browser-use (https://github.com/browser-use/browser-use), has just raised $17 million in seed funding. We've previously explored how LLMs interact with browsers, such as with OpenAI's Operator and Anthropic's Computer use. In today's episode, we'll take a closer look at browser-use, diving deep into how this tool can enable these capabilities, highlighting its strengths, exploring its limitations, and uncovering the exciting opportunities ahead.

High-quality labeled data is crucial for training successful AI models, but collecting it at scale has always been a major challenge. In this episode, we’ll decode the story of Scale AI, one of the standout companies helping businesses annotate their data. We’ll explore its origins, how it scaled to work with hundreds of thousands of human labelers, and the challenges it continues to face.

In this episode, we dive into Deep Research, an agentic AI system that harnesses LLMs to automatically search the web, gather relevant information, and synthesize comprehensive research reports. We'll explore how these systems work behind the scenes, key vendors in this space, and discuss practical applications and use cases.

Last year, cybersecurity firm Wiz turned down a $23 billion acquisition offer from Google. Now, it's on track to double its annual recurring revenue to $1 billion by 2025. In this episode, we'll explore the story behind Wiz's rapid growth, dive into the technology powering its products, and uncover the unique strategies that help Wiz stand out in the fiercely competitive cloud security market.

If you'd like to know more about Wiz, here are the references used for the research:
https://www.wiz.io/blog/how-wiz-code-was-built-with-developers-in-mind

https://www.youtube.com/watch?v=ueM8XxkUSFI&t=278s

ElevenLabs is a pioneering AI company specializing in natural-sounding speech synthesis software. Founded in 2022, it has grown rapidly, reaching $100 million ARR in just 2 years. More recently, Spotify just opened up the support for AI generated audiobooks (https://newsroom.spotify.com/2025-02-20/spotify-opens-up-support-for-elevenlabs-audiobook-content/).

In this episode, we discuss the technology behind realistic speech synthesis, their unique approach and business model.

If you want to know more about ElevenLabs or the technology, you can refer to the following materials used in the research:
- https://github.com/neonbjb/tortoise-tts
- https://elevenlabs.io/blog...