"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"

Takeaways

Matt 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.

Chapters

00:00 Introduction and New Beginnings
02:28 Transitioning to Cloudflare and Team Dynamics
05:25 Exploring the Cloudflare Office and Culture
08:20 The Evolution of Gitpod to OWNr
10:58 Understanding Remote Development Environments
14:04 The Role of Agents in Software Development
16:50 Inner Source and Collaboration in Large Companies
19:51 The Future of Remote Development and AI Agents
22:33 Harnesses and Their Role in Development
26:07 The Power of Interview Questions
28:14 Innovative Business Models in AI
31:12 The Evolution of AI Tools
34:07 Optimizing Code for Agents
37:56 Guardrails in AI Development
46:24 Personal AI Workflows and Productivity

"It's always the best."

Chapters

00:00 Introduction and Setting the Stage
02:34 Luke's Journey to ElevenLabs
04:54 Growth Strategies and Hiring at ElevenLabs
06:58 Understanding Growth Teams
09:50 The Role of Product in Growth
12:33 Working Dynamics at ElevenLabs
15:15 Use Cases for ElevenLabs' Platforms
18:04 Voice Agents and Market Opportunities
20:59 Product-Led Growth Strategies
24:29 The Power of Product-Led Growth (PLG) in Customer Acquisition
26:19 Understanding On-Prem Solutions and Their Spectrum
28:05 Innovations in AI Deployment with AWS Bedrock
31:14 Navigating the Music Industry: Partnerships and Rights
34:54 The Horizontal Approach: Integrating Diverse Product Lines
38:26 Imagining Future Workflows with AI Assistants
47:30 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"

Takeaways

The 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.

Chapters

00:00 Podcast Jingle and Creative Process
02:34 Introduction to Torsten Ball and His Work
05:06 The Journey into Programming
07:44 The Impact of AI on Content Creation
10:23 Cultural Phenomena in Media
13:23 Exploring Global Trends in Social Media
17:59 The Evolution of Social Media Platforms
18:21 Journey into Programming and Early Interests
20:53 Rediscovering Programming and the Joy of Learning
24:45 Career Path and the Impact of the Internet
27:48 Transitioning to Remote Work and Personal Life
29:11 The Emoji Debate: A Personal Perspective
35:24 The Importance of Training in Tech
42:26 The Future of Software Development in an AI World
44:23 Evaluating Software Engineering and Product Quality
45:35 The Shift in Engineering Skills and Training Needs
47:29 Navigating the Challenges of AI in Software Development
51:11 Structuring Codebases for AI Integration
55:48 Leveraging AI for Efficient Debugging and Development
01:00:33 Creating Feedback Loops with AI Tools
01:05:20 The Future of AI in Software Engineering

Chapters

00:00 Introduction and Context of the Podcast
07:23 Updates on SimplePoll and AI Integration
18:14 Telemetry and Continuous Improvement in Development
24:30 Acquisitions and Industry Insights
32:37 Discussion on GPT-5 and Future of AI Models
39:13 Consumer Experience with AI Models
48:29 The Future of AI Content Verification
58:17 The Impact of AI on Content Creation
01:05:39 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.

Chapters

00:00 Introduction and Setup
02:38 Meeting Danilo Leal: Background and Career
05:33 The Evolution of Design and Engineering
08:17 Transitioning to Material UI
11:13 The Role of AI in Design
14:10 Prototyping and Development Process
16:58 The Future of Design and Technology
28:28 The Comfort of Familiarity in Development
29:18 Debugging Strategies in Zed
30:09 Building Tools and Prototyping in Zed
33:13 The Evolution of GPUI and Design Process
35:16 Future of GPUI and General Purpose Applications
38:11 AI's Role in Enhancing Workflow
40:44 Parallelization and Task Management in Development
43:47 The Future of Editors and Asynchronous Tasks
48:32 The Role of Code in the Age of AI
54:28 Integrating Context and Knowledge in Codebases

TLDR

• Matt 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.

Chapters

00:00 Cactus Conversations and Personal Anecdotes
01:00 Fishing Adventures and Culinary Delights
03:32 Water Sports and Outdoor Activities
06:16 Exploring Sub-Agents and Context Management
08:03 A2A Protocols and Future of Agent Interactions
10:54 Challenges with Claude Code and Testing Workflows
13:17 Sub-Agents: Benefits and Limitations
17:59 Navigating Code Actions and Permissions
20:21 Building Personalized AI Tools for Code Review
24:05 Challenges in Prompt Writing and Feedback Loops
30:11 Innovations in UI Rendering and Client-Server Interaction
37:12 The Future of Advertising and Technology
39:24 Living in the Future: Startup Ambitions
40:36 Anticipating Future Needs in Technology
42:57 The Role of Agents and Micro Payments
45:25 The Art of Prompting AI
49:34 The Future of Work and Communication
51:47 Time Zones and Programming Challenges

News
The pod has twitter/x - https://x.com/badagentpod
Wil’s new brand - tritanclub.com

Links

Boris 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.html

Emmett Shear’s Tweet Thread on AI use and speed - https://x.com/eshear/status/1944867426635800865

TLDR

• The 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 London

The Evolution of React and Sunil's Contributions

Transition to Cloudflare and the Concept of Durable Objects

Building PartyKit and Its Impact

The Role of AI Agents and Their Integration

Challenges and Opportunities in the Tech Landscape

Exploring Durable Objects in Cloudflare

Challenges of Real-Time Applications

Local First Programming vs. Durable Objects

The Simplicity of Pull-Based Models

AI and the Future of Coding

AI's Impact on Professions

The Future of AI and AGI

Innovations in Cloudflare's Sandbox

The Concept of Self-Building Applications

LLMs 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."

Takeaways

Integrations 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.

Summary

In 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.

Chapters

00:00 Introduction and Admin Chit-Chat
02:55 Exploring Integrations and Podcast Formats
05:50 Cloudflare Containers and Their Use Cases
08:43 Cold Starts and Performance Considerations
11:41 Pricing and Competitive Landscape of Cloud Services
14:41 Integrations and AI Tools Development
17:38 Claude Code SDK and Future Developments
26:10 Navigating API Challenges
27:47 Customizing Tool Calls for Integration
28:56 The Future of API Integration
30:09 Harnessing the Power of Code SDKs
33:49 The Importance of Harnesses in Development
34:54 Building Better Tools with Specifications
37:11 The Role of Type Safety in Code Generation
40:14 Exploring the AISDK and Its Benefits
42:16 Reflections on Work Off-Sites
48:38 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/cortex

Claude Code in Github - https://github.com/anthropics/claude-code-action

Shippie - https://github.com/mattzcarey/shippie

Amp - https://sourcegraph.com/amp
Claude Code Github action system prompt - https://gist.github.com/mattzcarey/dd52a8e1df710c98b44072de46dcc09a

cursor-rules-to-claude - https://github.com/StackOneHQ/cursor-rules-to-claude

Gitlip - https://www.gitlip.com/

Durable Object file system (DOFS) - https://www.npmjs.com/package/dofs

Muppet (MCP) - https://www.muppet.dev/docs

Supermemory MCP blog - https://supermemory.ai/blog/the-ux-and-technicalities-of-awesome-mcps/

Zapier Agents - https://zapier.com/agents

SPARQL - https://www.ontotext.com/knowledgehub/fundamentals/what-is-sparql/

Vibe-rules - https://www.npmjs.com/package/vibe-rules

Simple 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.ai

Shippie - autonomous code review and QA
thinkynote.com - a word processor with AI
SimplePoll - the easiest way to do polls on Slack
Kolo - 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.