The Man Who Mastered Claude: The Fastest Way To Use AI To Save Time in Business & Life (No Coding)

Summary

The episode presents Claude as a delegated assistant that can work across a person’s computer and connected services—not just answer one-off questions. Matt Ebso recommends treating it like a new employee: explain the goal and context, provide tools and boundaries, let it work, then review the result and give feedback.

Key Concepts and Guidance

• Delegate, automate, eliminate: First remove unnecessary tasks, automate work that can run reliably on its own, and delegate what remains.
• Structure prompts: Matt recommends giving Claude context, a task, and an output format. For recurring work, he adds a research phase to develop the process, then turns the refined process into reusable instructions, or a “skill.”
• Provide enough context: Don’t assume Claude will remember preferences or infer requirements. Matt describes each interaction as if an intelligent worker has to reread the available context. Detailed background and feedback help improve later runs.
• Set autonomy and limits: For low-risk tasks, users can explicitly authorize Claude to act without asking for approval each time. Matt also describes boundaries, such as not creating accounts or entering payment details. He advises checking scheduled-task results and correcting them over time.
• Choose the right Claude product:
  • Chat is like a consultant for questions, research, and reports.
  • Cowork is more like a project assistant or manager that can use connected services and project folders, produce files, and handle ongoing tasks.
  • Claude Code is more like a developer: it can build apps, websites, and technical automations. Matt says it can also be useful for non-developers when the desired solution is durable and should work without a Claude session running.
• Connectors and model choice: Claude can use connected services such as Google Workspace, Slack, Microsoft 365, Notion, Asana, and QuickBooks. Matt recommends more capable models for novel, complex research and less expensive models for established, routine procedures.

Examples and Demonstrations

• Reported savings: Matt says Claude identified several ways to save him about $20,000 in a year. One was a tax-filing opportunity worth about $18,000, which he checked with his accountant before acting on. Another was spotting an unusually high electricity rate. These are Matt’s reported examples, not independently verified figures.
• Recurring personal tasks: In a live demonstration, Matt asks Claude to research barbers in Hackney, compare their reviews and suitability, and arrange a haircut every two months. Claude creates a process and shortlist, checks his calendar, and avoids bookings requiring advance payment under the stated limits.
• Apartment search: Matt describes using Claude to find rentals across housing platforms, organize listings in a spreadsheet, score them against his and his wife’s criteria, and update learned preferences based on their feedback. The intended next step was drafting personalized messages to agents.
• Business research and marketing: Matt uses Claude for research in his software education work. For Sound Toggle, he says Claude monitors his site and competitors, looks for relevant topics, analyzes the product’s code repository, and drafts articles to improve visibility in search and AI-generated recommendations.
• Podcast guest research: Claude analyzes a show’s public catalog and performance, establishes guest-selection criteria, researches candidates, and prepares a personalized outreach email. In the demonstrated report, it analyzed 198 videos and 155 release dates, identified patterns in stronger episodes, scored prospective guests, and suggested additional opportunities. The host provides feedback so later searches better reflect the show’s preferences.
• Custom meal-planning app: Matt had Claude Code build a personal meal-planning tool around dietary restrictions and preferences. It creates weekly meal plans, recipes, and shopping lists; sends approved items to a grocery app; and uses ratings to improve future suggestions. It runs as a progressive web app (PWA)—a website that can be added to a phone’s home screen and used much like an app. Matt says it was designed to reduce the 12–16 weekly hours he and his wife estimated they spent on food-related planning and preparation.
• Zapier and HeyGen: Sponsored segments show Zapier creating a workflow that sends guest-booking details from a pre-interview into Slack, and HeyGen turning a short recording into an on-demand video avatar. These are product demonstrations, separate from Matt’s main Claude examples.

Practical Takeaway

Matt’s suggested first step is to dictate your biggest recurring frustrations in life or work to Claude, with as much relevant context as possible. Ask it to identify what could be handed off and what might save an hour or two each week. Then have it research a repeatable approach, test the result, provide feedback, and schedule the improved process.

Main Speakers and Sources

• Calum Johnson — Host of The Calum Johnson Show.
• Matt Ebso — Guest, associated with Medics Media and Sound Toggle; shares the Claude workflows and demonstrations.
• Sponsored product sources: Zapier and HeyGen.