Tim Desoto, a 49-year-old founder and CEO based in San Francisco, launched his AI-powered shopping platform, Goodlife, in late 2024 without a technical background. Since then, he’s learned firsthand where AI excels—and where it falls short. While AI has become a cornerstone of his daily operations, Desoto emphasizes that it’s not a one-size-fits-all solution. He’s learned to balance automation with human insight, especially when it comes to strategy, judgment, and complex problem-solving. Desoto relies heavily on his network to stay informed about the latest AI tools. Attending meetups and developer conferences in San Francisco, he listens closely to what others in the community are using. Recent trends he’s noticed include the rise of agentic workflows—systems where AI agents can autonomously perform tasks. OpenClaw, an open-source autonomous agent project, gained attention, while Moltbook made agent interactions visible in a social context. Claude Cowork is also emerging as a popular choice for teams seeking secure, enterprise-ready agent workflows. For his business, Desoto uses a mix of AI tools across his workflow. His paid stack includes advanced models like Claude Max, Gemini Ultra, and ChatGPT Business, along with productivity tools such as Cursor, Figma Make, Notion AI, Superhuman Ask AI, and Lovable, an AI website builder. He’s particularly impressed with Gemini’s latest image models, noting faster performance, more stable reasoning, and stronger multimodal capabilities—especially in generating consistent visuals during edits. He’s also seen improvements in real-time information responses. He structures his AI use through what he calls an “AI conveyor belt” exercise. He starts with a written prompt, then shifts to multimodal inputs, speaking aloud to the model and engaging in back-and-forth dialogue. He values when models challenge his assumptions, as many tend to be overly agreeable. After getting an initial output, he runs the same task through multiple models to compare results. For long-form analysis and structured insights, he prefers Claude and Gemini, especially Gemini’s inline source linking for fact-checking. For formal writing and reasoning, he leans on ChatGPT and Claude. For creative exploration and visual ideation, he uses both Gemini and ChatGPT to generate early concepts and mockups. This process can take anywhere from 15 minutes to several days, depending on the complexity of the decision. Despite AI’s power, Desoto eventually needed human developers. Early on, he “vibe coded” the alpha version of his product, but found that 30% to 40% of the output was flawed. He spent hours debugging across multiple screens, using AI to debug AI, until he reached about 95% confidence. Once he brought in developers, the product began advancing at a much faster, more reliable, and scalable pace. He now recognizes that while AI can do incredible things, technical expertise amplifies its potential in ways a non-technical founder simply can’t replicate alone. Above all, Desoto believes the human perspective remains irreplaceable. He’s moved from informal mentorship to formal advisors—people with deep experience who help him spot blind spots, identify opportunities, and connect with potential partners. These relationships have been invaluable. He’s learned that AI can generate options and accelerate work, but choosing the right path still demands human judgment, strategic thinking, and taste. The most important role AI plays, he says, isn’t replacing humans—it’s helping them think better.