Agents at work

In this conversation, Jordi Montes and Tereza Tizkova dive deep into the challenges and opportunities around running AI-generated code safely and efficiently. Tereza explains how their E2B provides essential cloud sandbox infrastructure designed specifically for AI agents.The discussion explores why AI-generated code must be executed in secure, isolated environments and how E2B’s solution leverages micro virtual machines (micro VMs) to offer enhanced security, scalability, and speed. She highlights real-world applications like web app generation, deep research, data analysis, and computer-use agents,.They also address critical questions about open-source strategy, evolving LLM capabilities, industry benchmarks, and the exciting growth of AI agents in production settings.00:00 Introduction to E2B and AI Agents03:10 The Infrastructure Behind AI Agents05:48 The Evolution of E2B and Its Market Impact09:00 Customer Insights and Use Cases12:01 The Future of AI Agents and E2B's Role14:52 Open Source and Community Engagement18:02 Challenges and Opportunities in AI Development21:06 The Role of Data in AI Agents23:48 Exciting Developments and Future Directions27:13 Getting Started with E2B33:01 AI Hackathon in Prague: A New Era for Europe34:11 The European AI Landscape: Challenges and Opportunities35:44 Innovative Projects: Surprising Uses of E2B37:03 Understanding LLMs: Code Generation and Reasoning39:00 Versioning and Iteration: Enhancing Code Execution41:04 The Future of AI Infrastructure: Pricing and Data Insights45:32 The Evolution of AI Agents: What Lies Ahead51:00 Defining AI Agents: Capabilities and Future Trends

In this episode, Jordi Montes and Johno Whitaker (https://answer.ai) discuss the evolving landscape of AI agents, exploring definitions, the impact of generative AI, and how AI development is becoming more accessible. Answer.AI is a new kind of AI R&D lab which creates practical end-user products based on foundational research breakthroughs. You can think about it as Xerox Labs for the AI era.They delve into notebook-driven development, integration of AI in software workflows, and debate whether the future will feature powerful single models or cooperative networks of smaller agents. The conversation emphasizes empowering users through thoughtful interaction design and innovative frameworks like LLMs.txt and FastHTML, highlighting AI's potential to enhance human creativity and cautioning against over-reliance on automation.

In this episode of Agents at Work, Jordi Montes interviews Aiswarya from Entelligence, a company building AI engineering agents to help large engineering teams reduce overhead and focus on product development. Aiswarya explains how Entelligence's tools assist with code reviews, documentation, and team coordination. She explains how their AI agents can catch issues in code, keep documentation continuously updated, and provide insights for managers about team productivity. She shares insights about the limitations of current AI coding models and how their product complements code generation tools by serving as an "auditor" that catches potential bugs and enforces best practices. The discussion highlights how AI tools are transforming engineering team dynamics, making information more transparent across organizations, and changing how engineering performance is measured in the age of AI-assisted coding.

In this episode, Jordi Montes speaks with Arjun Chintapalli and Bhavani Kalisetty, co-founders of Retriever (https://rtrvr.ai), about the evolution of browser AI copilots. Arjun shares his background in federated learning at Google before founding RTRVR in December. Browser extension approach to AI assistance represents a significant advancement over traditional automation methods. By interpreting DOM representations rather than screenshots, this technology achieves greater speed, accuracy, and reliability compared to vision-based alternatives. The conversation explores how browser-based AI agents can simultaneously work across multiple tabs, extract structured data from websites, and integrate directly with productivity tools. All at a fraction of traditional web scraping costs. The technical advantages of operating within the user's browser environment include bypassing anti-bot systems, utilizing existing user credentials, and maintaining minimal performance overhead. They also discuss their vision for a workflow exchange marketplace where users can share automation recipes.

In this episode, Jordi Montes and David from Aomni discuss the evolving landscape of AI agents, their definitions, and their implications for software interaction and sales operations. David and his projects went viral multiple times the last one being an open source version of Deep Research. His company, Aomni, enables B2B sales reps save more than 7 hours per week on prospect research.David shares insights on building autonomous agents for revenue teams, the challenges of current software stacks, and the future of AI in enterprise sales. We talk about the role of tool calling in AI development and the potential for integrating coding with AI capabilities. We cannot stress enough importance of not underestimating their capabilities. The dialogue also touches on user interaction, the role of context and memory in AI applications, and the challenges of service discovery and performance evaluation in agent-based systems.

In this episode of Agents at Work, Jordi Montes interviews Christopher David, founder and CEO of OpenAgents, discussing the evolution of AI agents, their functionality, and the future of coding agents. Christopher shares insights on the development of ONYX, a mobile app for coding agents, and the importance of transparency and trust in AI development. The conversation also touches on the impact of new AI models, the role of interfaces, and the vision for a Bitcoinized economy where agents can operate autonomously.