DecodeFM

Anthropic just published an interesting report on how people are using AI in the real world by analyzing millions of real conversions on their Claude.ai platform. In this episode, we'll discuss some of the key findings in the report, for example, who's using AI the most (hint: software developers) and how they are using it. 

If you'd like to read the full report, it is here:
  https://assets.anthropic.com/m/2e23255f1e84ca97/original/Economic_Tasks_AI_Paper.pdf

OpenAI recently introduced Operator (https://openai.com/index/introducing-operator/), an AI agent that can use its own browser to perform tasks for you. Anthropic has a similar product called computer use (https://www.anthropic.com/news/3-5-models-and-computer-use), and there are open source projects like browser-use (https://github.com/browser-use/browser-use). In this episode, we take a look at the technology behind the so-called Computer Using Agents (CUA), their use cases, limitations and the challenges ahead.

In this episode, we dive into DeepSeek, the hottest AI company making waves in the industry right now. We'll explore the company's origin story, discuss what's unique about its v3 and R1 models, and why DeepSeek is able to train these models at a fraction of the cost compared to traditional methods.

If you are interested in learning more about DeepSeek and its models, we've included some references to satisfy your curious mind:
- DeepSeek v3 Technical Report: https://arxiv.org/pdf/2412.19437
- DeepSeek R1: https://github.com/deepseek-ai/DeepSeek-R1
- SemiAnalysis' analysis: https://semianalysis.com/2025/01/31/deepseek-debates/
- Founder's interview: https://www.lesswrong.com/posts/two-interviews-with-the-founder-of-deepseek

In this episode, let's talk about Replicate, a company on a mission to make running AI models as simple as making an API call. We'll discuss its origins, the business model, its strong ties to the open source movement, and some potential challenges.

In this episode, we look at another AI coding assistant - Lovable. It got so popular that it was shut down by github early this month  (https://lovable.dev/news/github-incident-2025-01-02). What can it do and how does it compare to other AI coding assistants?

In this episode, we discuss Devin AI, a $2B company developing an AI-powered coding assistant, a.k.a. AI Software Engineer. Its demos are impressive, but is it real or just a toy? Can it handle real-world projects? How does it compare to other
AI coding assistants? We'll also discuss how AI coding assistants might reshape the future of software engineering and what it means for human developers.

Gong is transforming sales with AI-powered revenue intelligence, analyzing sales conversations to help teams make smarter decisions. It's also used for coaching salespeople. In this episode, we’ll explore how Gong works, its alternatives, user feedback, and what’s next for this innovative company.

Welcome to 2025!

In this episode, we’ll discuss Anthropic and its rivalry with OpenAI. Known for its cautious approach to AI, how does Anthropic balance the need to move fast as a startup with its focus on AI safety? What are they focusing on in both business and technology as they navigate the rapidly evolving AI landscape?

Happy New Year!

In this last episode of 2024, we'll look at Rippling, the Swiss Army knife for HR. Valued at $13 billion, it offers a range of tools to help companies manage their workforce. It has a unique founder-led startup culture. In fact, it has hired 150(!) former startup founders. In this episode, we’ll discuss Rippling's culture, its business model, how it navigates challenges like the SVB shutdown, and its plans for the future.

Revolut first started as a company in 2015 to provide no-fee transactions for international travelers. Since then, it has grown into an all-in-one fintech powerhouse, with over $2 billion in revenue and $45 billion in valuation. In this episode, we tell the story behind its rapid growth, what it's like to work there, and how the company plans to stay ahead of competitions?