1011: tmux + Terminal Maxxing with Ben Vinegar

The episode begins with hosts Scott and Wes welcoming Ben Vinegar, former GM of Syntax and co-founder of Modem.dev. They discuss the excitement about Ben's work on AI project management and his innovative approach to terminal workflows. There’s a lighthearted tone as they mention how frequently Ben appears on the podcast and share some humorous comments about AI tools and coding environments.

Ben introduces Modem, a platform launched about a year ago that focuses on AI-powered product management. It helps streamline non-coding tasks like synthesizing customer feedback, managing ticket backlogs, and following up with users. The core idea is to leverage AI coding and automation to make software development faster and more efficient, with a focus on curated, reviewed code generation.

The conversation dives deep into terminal-centric productivity, especially Ben’s use of remote environments, SSH, tmux, and the temptation to move all workflows into the terminal. Ben emphasizes the benefits of tools like tmux for maintaining persistent workflows across different locations and devices. He also discusses integrating AI into terminal workflows, including commanding agents through CLI and embedding AI into environment management for seamless development experiences.

Ben demonstrates several tools he has developed or uses, such as Hunk, a terminal code review tool with GitHub stars, and TermDraw, a terminal diagramming application. He explains how these tools help visualize code, review changes, and facilitate communication without leaving the terminal. There is also mention of other utilities like fresh for terminal editing, and the importance of responsive, web-like GUIs within terminal applications.

The hosts and Ben discuss the proliferation of open-source contributions, PRs, and how to determine if feature requests are genuine. Ben shares his experience with building tools like Hunk and TermDraw, and the importance of tailored solutions like Pi, a minimal CLI environment, for specific workflows. The topic touches on managing open source quality and the importance of structured, easy-to-use tools for developers.

A critical discussion revolves around the security risks of 'dangerous mode' AI agents, with concerns about destructive actions like deleting data or dropping databases. Ben advocates for strict gates and safeguards, emphasizing that systems should be designed to prevent accidental or malicious damage. The debate highlights the tension between powerful automation and necessary safety measures.

Ben describes how to handle images and videos within terminal environments, including rendering PNGs with Kitty over SSH and using buildup automation for media previews. He mentions tools for visualizing code and diagrams directly in terminal, leveraging new WebAssembly-based technologies, and the potential of GUIs within terminal applications. This reflects a push toward richer multimedia capabilities in command-line interfaces.

Ben discusses his podcast, 'State of Agentic Coding,’ which explores the integration of AI and development tools. He shares his process for editing podcasts, including using open-source tools like PodGuy that streamline transcript generation, applying AI for editing decisions, and managing the production cycle. This segment emphasizes the importance of documentation and community sharing around AI-assisted development.

Ben explains how Modem serves as a product intelligence platform that aggregates user feedback from various channels to give teams real-time insights. His focus is on making feedback actionable and automating responses, such as creating tickets from social signals or internal feedback. The vision is to build AI-powered agents that work within the workflow to improve product quality and responsiveness across different domains.