In this episode, Colleen Schnettler shares how she built a custom AI-powered tool that turns her voice notes into LinkedIn posts, streamlining her content creation process and tailoring it to her workflow. She emphasizes the importance of building personalized tools, using AI for rapid development while maintaining hands-on quality control, and organizing content with clear frameworks for effective marketing.
In this episode of Builder Stories, host Brian Castle interviews Colleen Schnettler, a Ruby on Rails developer and entrepreneur, about how she leveraged AI to build her own marketing tools. Frustrated by the generic and robotic nature of existing AI-powered LinkedIn automation products, Colleen decided to create a custom solution tailored to her workflow. In just two days, she built a system that converts her voice notes into LinkedIn posts, allowing her to capture ideas throughout the day and quickly generate draft content for her consulting business.
Colleen’s process begins with recording voice notes using tools like Super Whisper or Whisper Flow. Her custom app monitors the folder where these transcripts are saved, reads the content, and automatically generates LinkedIn posts. This workflow enables her to easily capture spontaneous insights and turn them into social media content without the friction of manual posting. She emphasizes that while the AI-generated drafts may not be perfect, they serve as valuable starting points for further editing and refinement.
To build and iterate on her tool, Colleen uses the Compound Engineering framework from Every, which provides structured planning and execution through a set of slash commands in Claude Code. She demonstrates how she creates detailed plans for significant features, such as adding a chat interface for editing LinkedIn posts, and then reviews the AI-generated code before merging it. Colleen prefers a spec-driven approach for larger features, using high-level plans as guardrails while trusting the AI to handle most of the implementation.
Despite relying heavily on AI for coding, Colleen maintains a hands-on approach to quality assurance. She runs multiple AI agents in parallel, switches between Claude and GPT-5 when necessary, and always manually tests new features in the browser to ensure the app feels right. She notes that while AI can automate much of the development process, there is still an intuitive aspect to software that requires human judgment and review.
Colleen also shares her personal strategy for LinkedIn content, organizing posts into categories like “teach me,” “help me,” “inspire me,” and “show me.” She plans to use this framework not only for her own posts but also to help clients with their LinkedIn marketing. The key takeaways from her experience are: build your own tools to fit your needs, keep specifications light but purposeful, and always trust your own sense of quality and usability. The episode concludes with an invitation to access the full conversation and additional resources through the Builder Methods Pro community.