DiscoverYou've Been a Bad Agent
You've Been a Bad Agent
Claim Ownership

You've Been a Bad Agent

Author: Wilhelm Klopp & Matt Carey

Subscribed: 3Played: 21
Share

Description

Wil and Matt discuss tech, startups, and building really cool things with AI. Sometimes joined by (actual expert) friends.
18 Episodes
Reverse
Are we in a moment? The OpenClaw/Cloudbot autonomous agent craze explainedWhy everyone suddenly bought Mac Minis (and why Matt refuses to)Moltbook: Reddit for agents where bots questions their mortalityFiles vs SQLite for agent memory - what actually works for long-term storageMCP's future: stateless tools, elicitation, and a world with no human in the loopWalled gardens vs open internet - will APIs open up for agents to roam free?Models getting better at "just keep going" - the metric Anthropic optimized forThe massive rift between frontier AI users and "stochastic parrots" skepticsBlacksmith CI: Wilhelm's hot tip for faster GitHub ActionsClosing question: If Moltbook is Reddit for agents, what is GitHub for agents?Architecture twitter article about Clawdbot: https://x.com/Hesamation/status/2017038553058857413
Matt's Big Move to LisbonCloudflare Code Mode: 2,500 Tools in 1,000 TokensCode Mode vs Bash vs SQLite: What Actually Works for Agents2025 Reflections: The Year of Claude Code and MCP2026 Preview: Files Are BackQuick Fire: Best Model, Best Lab, Biggest Surprises
CGPGrey Video on Yearly Themes
"chat to your mouth Claude and grow a new tooth""I think I would have like a huge ego now for sure""have you seen the blimp"TakeawaysAll of these people in this industry have suddenly become important at work.Their work suddenly became important to a lot of the world.If I had worked at Anthropic five years ago, I would have a huge ego now.The growth in the AI industry feels unevenly distributed.The current time feels strange due to the rapid changes in AI.Ego can be a significant factor in how individuals perceive their roles.The importance of work can influence personal identity.The AI landscape is evolving in unexpected ways.Understanding the psychological impact of industry changes is crucial.The conversation reflects on the novelty of the current AI growth.
"a brief history of a very specific building in London.""there's a lot of room to optimize your code base for an agent.""70 % of our code is written by an agent.""The place I see where humans are still add tons of value….""spending 10, 15 minutes on a prompt is not unusual.""I’m not sure I wanna wear a muzzle"TakeawaysMatt recently joined Ona, previously known as Gitpod, and is excited about the developments there.The transition from Gitpod to Ona reflects a broader evolution in remote development environments.Remote development environments can significantly enhance productivity and onboarding for new engineers.AI agents are becoming integral to software development, allowing for automated code writing and testing.The importance of inner source practices in large organizations can facilitate collaboration and innovation.Amp Free represents a shift in how AI tools can be monetized and accessed by users.Guardrails are essential for ensuring that AI agents operate within safe and defined parameters.The balance between security and productivity is a constant challenge in software engineering.Personal AI workflows can streamline tasks and enhance efficiency in development processes.The future of AI tools will likely involve more integration with existing workflows and systems.Chapters00:00 Introduction and New Beginnings02:28 Transitioning to Cloudflare and Team Dynamics05:25 Exploring the Cloudflare Office and Culture08:20 The Evolution of Gitpod to OWNr10:58 Understanding Remote Development Environments14:04 The Role of Agents in Software Development16:50 Inner Source and Collaboration in Large Companies19:51 The Future of Remote Development and AI Agents22:33 Harnesses and Their Role in Development26:07 The Power of Interview Questions28:14 Innovative Business Models in AI31:12 The Evolution of AI Tools34:07 Optimizing Code for Agents37:56 Guardrails in AI Development46:24 Personal AI Workflows and Productivity
"It's always the best."Chapters00:00 Introduction and Setting the Stage02:34 Luke's Journey to ElevenLabs04:54 Growth Strategies and Hiring at ElevenLabs06:58 Understanding Growth Teams09:50 The Role of Product in Growth12:33 Working Dynamics at ElevenLabs15:15 Use Cases for ElevenLabs' Platforms18:04 Voice Agents and Market Opportunities20:59 Product-Led Growth Strategies24:29 The Power of Product-Led Growth (PLG) in Customer Acquisition26:19 Understanding On-Prem Solutions and Their Spectrum28:05 Innovations in AI Deployment with AWS Bedrock31:14 Navigating the Music Industry: Partnerships and Rights34:54 The Horizontal Approach: Integrating Diverse Product Lines38:26 Imagining Future Workflows with AI Assistants47:30 Personal AI Workflows: Tools and Automation
"Elderly German Landlord.. she would use these winky face emojis""Memes in blogs feel cheap""Dude.... you didn't even run this"TakeawaysThe podcast jingle was created using AI technology.Thorsten's journey into programming began in his teenage years.He transitioned from music to software engineering after realizing his passion for coding.The internet has significantly influenced Thorsten's life and career choices.AI-generated content is becoming a cultural phenomenon in Germany.Thorsten believes that some people can tolerate challenges better than others.He emphasizes the importance of continuous learning in programming.Thorsten's perspective on emojis highlights their humorous use but criticizes their overuse in UIs.The conversation reflects on the evolution of user experience in technology.Thorsten's career path showcases the diverse opportunities in the tech industry. The wink emoji is a cultural nuance in Germany.Training in tech is often on-the-job rather than formal.AI tools require engineers to rethink their approach to coding.Engineers need to understand the business context of their work.Dumb code can be generated quickly with AI, but structuring is key.Feedback loops are essential for effective AI integration.The software industry has unique training challenges compared to others.AI can assist in generating data visualization tools.Engineers should build tools that help them validate AI outputs.The landscape of software engineering is rapidly changing due to AI.Chapters00:00 Podcast Jingle and Creative Process02:34 Introduction to Torsten Ball and His Work05:06 The Journey into Programming07:44 The Impact of AI on Content Creation10:23 Cultural Phenomena in Media13:23 Exploring Global Trends in Social Media17:59 The Evolution of Social Media Platforms18:21 Journey into Programming and Early Interests20:53 Rediscovering Programming and the Joy of Learning24:45 Career Path and the Impact of the Internet27:48 Transitioning to Remote Work and Personal Life29:11 The Emoji Debate: A Personal Perspective35:24 The Importance of Training in Tech42:26 The Future of Software Development in an AI World44:23 Evaluating Software Engineering and Product Quality45:35 The Shift in Engineering Skills and Training Needs47:29 Navigating the Challenges of AI in Software Development51:11 Structuring Codebases for AI Integration55:48 Leveraging AI for Efficient Debugging and Development01:00:33 Creating Feedback Loops with AI Tools01:05:20 The Future of AI in Software Engineering
Chapters00:00 Introduction and Context of the Podcast07:23 Updates on SimplePoll and AI Integration18:14 Telemetry and Continuous Improvement in Development24:30 Acquisitions and Industry Insights32:37 Discussion on GPT-5 and Future of AI Models39:13 Consumer Experience with AI Models48:29 The Future of AI Content Verification58:17 The Impact of AI on Content Creation01:05:39 Community Engagement and Trends
TDLR: 10 episodes in and we are going full uncut conversation with Danilo Leal. One of the magician Design Engineers working on the Zed code editor. “I'll be making a jingle”“We prototype everything in React before Rust”Danilo Leal represents a new breed of designer and engineer.The role of design is becoming more approachable with technology.AI is changing the landscape of coding and design.Prototyping in React allows for rapid iteration and testing.Debugging in Zed involves traditional methods and new tools.Collaboration is key in modern development environments.The future of coding may involve multiple agents working in parallel.AI can enhance workflows but requires careful context management.The editor's role is evolving with the integration of AI tools.Design processes are becoming more integrated with coding practices.Chapters00:00 Introduction and Setup02:38 Meeting Danilo Leal: Background and Career05:33 The Evolution of Design and Engineering08:17 Transitioning to Material UI11:13 The Role of AI in Design14:10 Prototyping and Development Process16:58 The Future of Design and Technology28:28 The Comfort of Familiarity in Development29:18 Debugging Strategies in Zed30:09 Building Tools and Prototyping in Zed33:13 The Evolution of GPUI and Design Process35:16 Future of GPUI and General Purpose Applications38:11 AI's Role in Enhancing Workflow40:44 Parallelization and Task Management in Development43:47 The Future of Editors and Asynchronous Tasks48:32 The Role of Code in the Age of AI54:28 Integrating Context and Knowledge in Codebases
TLDRMatt caught whitebait for dinner during his outdoor adventures.Wing foiling is a new sport gaining popularity.Sub agents are necessary for exploring context windows in AI.A2A and MCP servers are the future of AI integration.AI-assisted code review can streamline the development process.Prompting techniques are evolving and require careful consideration.Living in the future means adapting to rapid technological changes.Time zones can create unexpected challenges in programming.The importance of personalized AI tooling is growing.Understanding the intricacies of AI models is crucial for effective use.Chapters00:00 Cactus Conversations and Personal Anecdotes01:00 Fishing Adventures and Culinary Delights03:32 Water Sports and Outdoor Activities06:16 Exploring Sub-Agents and Context Management08:03 A2A Protocols and Future of Agent Interactions10:54 Challenges with Claude Code and Testing Workflows13:17 Sub-Agents: Benefits and Limitations17:59 Navigating Code Actions and Permissions20:21 Building Personalized AI Tools for Code Review24:05 Challenges in Prompt Writing and Feedback Loops30:11 Innovations in UI Rendering and Client-Server Interaction37:12 The Future of Advertising and Technology39:24 Living in the Future: Startup Ambitions40:36 Anticipating Future Needs in Technology42:57 The Role of Agents and Micro Payments45:25 The Art of Prompting AI49:34 The Future of Work and Communication51:47 Time Zones and Programming Challenges
NewsThe pod has twitter/x - https://x.com/badagentpodWil’s new brand - tritanclub.comLinksBoris Tane post on Cloudflare DOs + Drizzle - https://boristane.com/blog/durable-objects-database-per-user/Experience with Claude Code - https://sankalp.bearblog.dev/my-claude-code-experience-after-2-weeks-of-usage/Focus on inputs not outputs: https://john-rush.com/posts/ai-20250701.htmlEmmett Shear’s Tweet Thread on AI use and speed - https://x.com/eshear/status/1944867426635800865TLDRThe experience level of developers affects their efficiency with AI tools.Proficiency in using AI coding agents requires practice and skill.Co-locating tests with code can improve the development process.Strict linting rules are essential for maintaining code quality.Different coding tasks require different approaches to AI assistance.The evolution of coding practices is influenced by new technologies.Understanding the limitations of AI tools is crucial for effective use.Developers need to adapt to new tools and workflows continuously.The future of development will likely involve more AI integration. Claude Code/TUI offers flexibility for running various applications.The evolution of coding agents is crucial for developers.Metrics of success in AI should focus on efficiency and user satisfaction.AI models should balance between asking questions and executing tasks.Personal experiences shape our understanding of AI tools.Innovations in coding tools can significantly enhance productivity.The future of AI involves deeper human interaction and understanding.Celebrating milestones is important for motivation and growth.Development workflows are being transformed by AI capabilities.Continuous learning and adaptation are key in the tech industry.
“UI is so over”“If all humans were perfect robots”"I just got stoned and did open source.""I rewrote their entire CLI."Sunil Pai's Backstory and Career Journey from India to LondonThe Evolution of React and Sunil's ContributionsTransition to Cloudflare and the Concept of Durable ObjectsBuilding PartyKit and Its ImpactThe Role of AI Agents and Their IntegrationChallenges and Opportunities in the Tech LandscapeExploring Durable Objects in CloudflareChallenges of Real-Time ApplicationsLocal First Programming vs. Durable ObjectsThe Simplicity of Pull-Based ModelsAI and the Future of CodingAI's Impact on ProfessionsThe Future of AI and AGIInnovations in Cloudflare's SandboxThe Concept of Self-Building ApplicationsLLMs are a mirror of an operators skill - https://ghuntley.com/mirrors/Sunil's blog - https://sunilpai.dev/
"Bullish on claude code""I found Soham in our ATS""these things have been like RLHF to fuck"Vibe Tunnel - https://vibetunnel.sh/Armin Ronacher on Simon Willison’s blog - https://simonwillison.net/tags/armin-ronacher/Amp by Sourcegraph - https://ampcode.com/Matt is finalizing his event for AI Demo Days.Juliette completed a challenging marathon with significant elevation.The tech news cycle is currently nuanced and interesting.Soham's job application saga has sparked widespread discussion.Roenacher's insights on Claude Code are valuable for developers.Git work trees can be beneficial but have their challenges.Environment variables can complicate development processes.Context engineering is crucial for effective AI interactions.Talent poaching is rampant in the tech industry right now.The competitive landscape allows companies to attract top talent easily. Cursor is refining its features to improve user experience.The competition between AWS and Cloudflare shapes developer tools.Claude Code is designed to work closely with AI models.AI can enhance productivity but requires careful oversight.Vibe Tunnel offers a novel way to manage terminal sessions.MCP servers need to return appropriate data for voice agents.Quality checks are essential for AI-generated code.Developers should prioritize and start small when building tools.The integration of tools should be user-driven, not server-driven.The evolution of MCP servers reflects the changing landscape of AI development.
"Context window… kaboom""Am I gonna get cancelled for this""United is not a good airline""Just make more money."TakeawaysIntegrations bring people together in meaningful ways.Video podcasts are gaining popularity among Gen Z audiences.Editing podcasts can be time-consuming but rewarding.Family events can provide a refreshing break from work.Cloudflare Containers offer new possibilities for developers.Pricing strategies in cloud services can impact user choices.AI integrations can simplify complex API interactions.Work off-sites can enhance team bonding and productivity.Travel experiences can shape preferences for airlines and destinations.Effective communication is key in team dynamics.SummaryIn this conversation, Matt Carey and Wilhelm Klopp discuss various topics ranging from the importance of integrations in technology and podcasting trends to personal updates about family events and travel experiences. They delve into the technical aspects of Cloudflare Containers and the Claude Code SDK, exploring their implications for developers and AI integrations. The discussion also touches on the dynamics of work off-sites and the impact of travel on airline preferences, emphasizing the importance of effective communication and team bonding.Chapters00:00 Introduction and Admin Chit-Chat02:55 Exploring Integrations and Podcast Formats05:50 Cloudflare Containers and Their Use Cases08:43 Cold Starts and Performance Considerations11:41 Pricing and Competitive Landscape of Cloud Services14:41 Integrations and AI Tools Development17:38 Claude Code SDK and Future Developments26:10 Navigating API Challenges27:47 Customizing Tool Calls for Integration28:56 The Future of API Integration30:09 Harnessing the Power of Code SDKs33:49 The Importance of Harnesses in Development34:54 Building Better Tools with Specifications37:11 The Role of Type Safety in Code Generation40:14 Exploring the AISDK and Its Benefits42:16 Reflections on Work Off-Sites48:38 Travel Experiences and Airline Preferences
"Cortex podcast is God tier""Claude Code is the best devtool this year""gotta structure your codebase for AI"Stuff we talked about:Cortex podcast - https://www.relay.fm/cortexClaude Code in Github - https://github.com/anthropics/claude-code-actionShippie - https://github.com/mattzcarey/shippieAmp - https://sourcegraph.com/ampClaude Code Github action system prompt - https://gist.github.com/mattzcarey/dd52a8e1df710c98b44072de46dcc09acursor-rules-to-claude - https://github.com/StackOneHQ/cursor-rules-to-claudeGitlip - https://www.gitlip.com/Durable Object file system (DOFS) - https://www.npmjs.com/package/dofsMuppet (MCP) - https://www.muppet.dev/docsSupermemory MCP blog - https://supermemory.ai/blog/the-ux-and-technicalities-of-awesome-mcps/Zapier Agents - https://zapier.com/agentsSPARQL - https://www.ontotext.com/knowledgehub/fundamentals/what-is-sparql/Vibe-rules - https://www.npmjs.com/package/vibe-rulesSimple Poll - https://simplepoll.rocks/Kolo - https://kolo.app/TLDR:Matt watched Yellowstone late at night and got hyped.Meta's acquisition of Scale raises questions about data distribution.Acquisitions in tech often lead to speculation and drama.Cortex podcast is highly recommended for productivity insights.Networking events can introduce you to new circles of people.Shippie has introduced new features like subagents.Subagents allow for parallel processing in coding tasks.Claude Code integration with GitHub simplifies development workflows.Documentation should include key concepts for better understanding.A product for coding agents should focus on code structure and tooling.
"We're always live.""I think evals are so hard.""AI engineering is dead."Stag do experiences can lead to unexpected personal updates.Consumer behavior in Las Vegas highlights the allure of gambling despite its downsides.Podcasting can encompass news, guest interviews, and personal stories.Traveling to Peru offers breathtaking experiences like hiking Machu Picchu.AI model releases are frequent and can impact development workflows.AI agents face challenges in complex coding tasks.New tech gadgets like the Garmin Tactics watch enhance outdoor activities.Fitness apps often lack user-friendly features for multi-sport activities.Conferences provide valuable networking opportunities in the tech industry.Deep research can uncover useful programming techniques and tools.AI Engineering is Dead? - Louis from Bloop.aiShippie - autonomous code review and QAthinkynote.com - a word processor with AISimplePoll - the easiest way to do polls on SlackKolo - useful things with runtime data in python
Lu, Wilhelm, and Matt discuss the evolution and features of tldraw, "a very good whiteboard". They explore the office culture, community engagement through demos, and the integration of AI technologies. The challenges faced in AI development and the importance of developer relations are also highlighted, emphasizing the need for effective communication and support for users.  There is chat about the unique challenges faced in SDK development in an era of LLMs, the creative environment of their office, the significance of physical workspaces, and the ongoing rumors surrounding acquisitions in the tech industry. They also discuss innovations in tldraw Computer, the launch of a triathlon brand, and the complexities of navigating the MCP landscape.
Wilhelm and Matt are starting a podcast together.They discuss the differences in humor between the UK and the US.Scented trash bags are surprisingly beneficial.Matt's Garmin watch has malfunctioned after years of use.Wilhelm is adjusting to life in San Francisco, noting its high cost of living.They plan to include guests in their podcast to enhance discussions.Matt runs AI demo days to showcase startup innovations.Cloudflare is praised for its deployment experience.Wilhelm shares his challenges with deploying a Python project.They reflect on the importance of genuine conversations in podcasting. Cloudflare offers out-of-the-box deployment solutions that simplify the process.Using modal for Python applications can significantly speed up deployment.Redwood's branding and recent name change reflect its evolution in the tech landscape.MCP servers are becoming essential for continuous training and fine-tuning of AI models.The popularity of MCP servers can be gauged through download statistics from NPM and PyPI.Building autonomous agents requires a clear understanding of feedback loops and context management.AI adoption varies widely among developers, with experienced engineers benefiting the most.Technical debt now includes the absence of structured rules files in codebases.The integration of AI tools into development workflows is still evolving.Maintaining a consistent coding pattern is crucial for leveraging AI effectively.
Comments 
loading