Title: Why you're wasting more time with AI than you realize Author: Natalie Lambert Published: 2026-09-08 Type: Newsletter — Prompt, Tinker, Innovate URL: https://genedge.co/newsletter/why-youre-wasting-more-time-with-ai-than-you-realize Excerpt: This kind of waste never shows up in your usage limit. It shows up in the time you spend re-explaining who you are and how you want to work, every time you open a new chat. Projects, Skills, and scheduled tasks each remove a different piece of that repetition—the question is one body of work, a reusable way of working, or work you shouldn't have to remember to trigger. --- Welcome to Prompt, Tinker, Innovate—my AI playground. Each edition gives you a hands-on experiment that shows how AI can sharpen your thinking, streamline your process, and power up your creative work. ## This week's playground: Stop retyping it, stop repeating the instructions, stop showing up to ask The last two editions were about token usage and limits—the dials that set your usage per message (/newsletter/why-youre-hitting-your-ai-limits-faster-than-you-expect), and the tools that set your usage before you've touched a dial (/newsletter/why-the-tool-you-choose-can-cost-more-than-the-question-you-ask). This week is about a different kind of waste, and it doesn't show up in your usage limit at all. It shows up in the time you spend re-explaining who you are, what you're working on, and how you want AI to work with you every single time you start a new conversation. Three features can remove different pieces of that repetition: Projects, Skills and custom assistants such as GPTs, Gems, and Copilot agents, and scheduled tasks. The distinction isn't simply context versus instructions—there's overlap. The better way to think about them is scope: one body of work, a reusable way of working, or work you no longer want to remember to trigger yourself. ## Projects Claude and ChatGPT both have a version of this. A Project is a persistent workspace, not a single conversation. Load it with the files, background, and standing instructions for one body of work, and conversations inside that Project can start with that foundation already in place. No re-pasting the same client brief, style guide, campaign background, or working rules every time you open a new chat. If you're a consultant, this might be one Project per client. If you write regularly, it could be one ongoing publication or campaign with its source material, editorial guidance, and background living in one place instead of getting re-pasted into every new chat. The payoff isn't necessarily a smarter answer—it's getting back the time you used to spend rebuilding the setup. Good for: an ongoing engagement with its own context and instructions—a client, a campaign, a recurring piece of work that always needs the same foundation. ## Skills, Custom GPTs, Gems, Copilot agents Don't confuse this with a Project. A Project packages what AI needs for one body of work. A Skill, Custom GPT, Gem, or agent packages a reusable task, role, or way of working that you may want to use across many different kinds of work. Each platform builds this a little differently. Claude Skills can contain detailed instructions, supporting files, and even code while keeping the description Claude uses to identify the right Skill intentionally short. Custom GPTs and Gems also let you package instructions, knowledge, and behaviors into reusable assistants. The mechanics vary, and they'll keep changing. The more useful question is whether the instructions belong to one specific piece of work or whether you want to reuse them again and again. If you catch yourself giving some version of the same instructions across different conversations—"write like this," "review it this way," "always check for X"—that's a good sign you should stop typing them manually and build a reusable assistant or Skill instead. Good for: a repeatable way of working you want available across different work—a review process, a voice, a research method, or a format you use again and again. ## Scheduled tasks Major AI platforms have a version of this. This is the same idea taken one step further. A Project reduces how often you rebuild the setup for a body of work. A Skill or custom assistant reduces how often you repeat the same working instructions. A scheduled task removes the need to remember to run the work at all—you give it the instructions once, add a cadence, and let it run on schedule. Once you've built the instruction—the weekly competitor scan, the Monday morning briefing, the recurring inbox triage—a scheduled task is what happens when you stop being the person who has to remember to trigger it. Claude's scheduled tasks in Cowork can run remotely even when your computer is asleep or the app is closed, although tasks that need access to local files or apps still depend on your machine. ChatGPT, Gemini, and Microsoft 365 Copilot's scheduled tasks can also run on their own schedules without you sitting there waiting to start them. Good for: recurring work with a fixed cadence you'd otherwise have to remember to trigger yourself—a status digest, a monitoring check, or a report you want generated on the same day every week. ## Your AI experiment: Find the thing you keep retyping 👉 Time to tinker: Look back at your last five or six conversations in any AI tool. Find something you gave it more than once—a client brief, a review process, a tone, a format, or a task you keep remembering to run. If it belongs to one ongoing body of work, put it in a Project. If it's a way of working you want to reuse across different work, build it into a Skill, Custom GPT, Gem, or agent. If it's something you keep coming back to run on a regular cadence, turn it into a scheduled task. 💡 Pro tip: Start with the one thing that annoys you most to repeat. That's usually the clearest sign of where better setup pays off fastest. My rule of thumb: when I find myself doing something for the third time, I force myself to set up the Skill. ## What did you discover? Did you find something you've been rebuilding for months without noticing? Or did moving it into a Project, Skill, or scheduled task change less than you expected? Tell me what surprised you. Until next time—keep tinkering, keep prompting, keep innovating. 📩 Not subscribed yet? Hit the button at the top. You won't want to miss what's next.