Vasek Mlejnsky, a visionary from E2B, joins to share insights on building secure cloud sandboxes for AI agents. He discusses the rapid growth of E2B and its adoption by major companies. The conversation dives into the unique challenges posed by early LLMs and the advantages of cloud environments for AI. Vasek highlights practical use cases like code execution and data analysis, while also addressing the shifting landscape of AI frameworks and billing models. His thoughts on future advancements and multi-modality in AI are particularly intriguing.
The transition to AI-driven sandboxes at E2B showcases a strategic shift to enhance developer engagement through interactive tools and environment flexibility.
E2B's significant user growth indicates a strong demand for AI agents capable of executing tasks like data analysis and code evaluation seamlessly.
The commitment to building a user-friendly interface highlights E2B's aim to accommodate developers with varying infrastructure expertise while fostering community partnerships.
Deep dives
Developer Experience and Interactive Tools
The discussion highlights the evolution of a project initially focused on enhancing developer experience through interactive documentation. The idea was to create an environment where developers could engage with tools via pre-made interactive guides instead of traditional reading and googling. One example mentioned is an interactive playground for Prisma that facilitated direct experimentation without complex setups. This initial work laid the groundwork for a more scalable infrastructure, transitioning into the current model of offering sandbox environments for experimentation.
Transition to AI and E2B Development
A pivotal moment occurred when the team shifted focus to the integration of AI tools, particularly after the release of GPT-3.5, which sparked their interest in automating development work. They explored building an AI agent that could manage backend integrations seamlessly. The success of their initial experiments, shared widely on social media, led them to open-source their work and form a community around their new venture, dubbed E2B, which stands for converting English instructions into actionable code. This strategic pivot allowed them to tap into growing interest in AI and developer tools.
Scaling and Market Fit
The transition to building a more robust platform involved addressing the need for environments where AI agents could operate effectively. They faced challenges in defining their go-to-market strategy while realizing that code interpreting was essential to attract a wider audience. Data from their user growth indicated a significant surge in the use of their sandboxes, showcasing a strong demand for environments that enhance AI capabilities. By engaging with users and iterating on feedback, the platform evolved to support tasks that ranged from AI-driven data analysis to crafting complex applications.
Navigating the AI Infrastructure Landscape
As the team reflects on their positioning within the broader AI infrastructure ecosystem, they emphasize the importance of remaining agnostic to various machine learning models. They also aim to construct a user-friendly environment that caters to developers without deep infrastructure knowledge. With aspirations to build a best-in-class platform for AI applications, they plan to enhance features that allow agents to autonomously manage sandboxes and run code. This forward-thinking approach showcases their commitment to enabling developers and AI agents to work seamlessly together.
Future Directions and Community Focus
Looking forward, the team envisions expanding their platform’s capabilities to support diverse use cases, including model training and evaluation. They are actively cultivating community partnerships, reminding that good relationships are vital in the AI space. Their efforts to tailor their services toward user feedback have laid a strong foundation for potential collaborations moving ahead, particularly with academic institutions. By shifting focus back to building user-friendly tools and environments for AI, they anticipate driving significant growth in both user engagement and product development.
Vasek Mlejnsky from E2B joins us today to talk about sandboxes for AI agents. In the last 2 years, E2B has grown from a handful of developers building on it to being used by ~50% of the Fortune 500 and generating millions of sandboxes each week for their customers. As the “death of chat completions” approaches, LLMs workflows and agents are relying more and more on tool usage and multi-modality.
The most common use cases for their sandboxes:
- Run data analysis and charting (like Perplexity)
- Execute arbitrary code generated by the model (like Manus does)
- Running evals on code generation (see LMArena Web)
- Doing reinforcement learning for code capabilities (like HuggingFace)
Timestamps:
00:00:00 Introductions 00:00:37 Origin of DevBook -> E2B 00:02:35 Early Experiments with GPT-3.5 and Building AI Agents 00:05:19 Building an Agent Cloud 00:07:27 Challenges of Building with Early LLMs 00:10:35 E2B Use Cases 00:13:52 E2B Growth vs Models Capabilities 00:15:03 The LLM Operating System (LLMOS) Landscape 00:20:12 Breakdown of JavaScript vs Python Usage on E2B 00:21:50 AI VMs vs Traditional Cloud 00:26:28 Technical Specifications of E2B Sandboxes 00:29:43 Usage-based billing infrastructure 00:34:08 Pricing AI on Value Delivered vs Token Usage 00:36:24 Forking, Checkpoints, and Parallel Execution in Sandboxes 00:39:18 Future Plans for Toolkit and Higher-Level Agent Frameworks 00:42:35 Limitations of Chat-Based Interfaces and the Future of Agents 00:44:00 MCPs and Remote Agent Capabilities 00:49:22 LLMs.txt, scrapers, and bad AI bots 00:53:00 Manus and Computer Use on E2B 00:55:03 E2B for RL with Hugging Face 00:56:58 E2B for Agent Evaluation on LMArena 00:58:12 Long-Term Vision: E2B as Full Lifecycle Infrastructure for LLMs 01:00:45 Future Plans for Hosting and Deployment of LLM-Generated Apps 01:01:15 Why E2B Moved to San Francisco 01:05:49 Open Roles and Hiring Plans at E2B
Remember Everything You Learn from Podcasts
Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.
