I used Naval Ravikant’s 'Leverage' rule with ChatGPT agents — and it cut my workload in half

Most people use AI like a smarter search engine. I’ve done plenty of that myself. But once you start using it with greater intention, you'll discover that you're missing the "billionaire mindset" of scaling yourself. Now that AGI is here, using AI like the top 1% can make a huge difference in output and productivity.

Entrepreneur and investor Naval Ravikant often talks about leverage, which is the ability to decouple your inputs from your outputs. In Naval’s world, there is "permissioned" leverage (labor and capital) and Permissionless Leverage (code and media).

Today, AI agents are the ultimate "permissionless" leverage. You don't need to hire staff or raise capital to build a team; you just need to build systems.

In the book, "The Almanack of Naval Ravikant: A Guide to Wealth and Happines," author Eric Jorgenson explores Ravikant's principles for building wealth and creating long-term happiness. After reading it, I couldn't help but ask ChatGPT
how to implement Ravikant's rules into my workflow. So for one week, I stopped "chatting" with AI and started deploying it as a "robot army" to handle my recurring work.

Here's a look at what happened and how to utilize this principle yourself.

What Naval Ravikant means by leverage

Leverage is about building systems that continue producing value after the initial setup. AI agents are especially good at this for workflows such as coding, media, automation or processes.

For instance, if you spend 30 minutes creating a workflow that saves you 30 minutes every day, that payoff compounds quickly. Over a month, you’ve reclaimed hours. Over a year, the number becomes significant. It's like having an assistant that you pay $20/month for (roughly the cost of ChatGPT Plus, Claude Pro or Perplexity Pro, all of which can handle automations).

That's the mindset I tested with five ChatGPT agents.

1. Inbox triage agent

Using ChatGPT's Gmail integration within the ChatGPT app hub, I asked ChatGPT to sort my emails into four specific buckets including urgent, important but not urgent, quick reply and ignore/archive.

From there, I asked ChatGPT to draft responses in my tone. Instead of mentally processing every email one by one, I started with a prioritized list. As someone with thousands of emails sitting idle in my inbox, the organization helped even more than using Gemini. I dictated the "buckets" each email went into and the AI utilized ChatGPT memory mode to know which bucket to put something in. For me, this method worked even better than Gemini's integration with Gmail.

Here's the prompt I used: You are my executive assistant. Review these emails and sort them into: Urgent, important but not urgent, ignore/archive, needs quick reply. Then draft concise replies in my tone.

2. Research agent

For topics I needed to understand quickly, I used a prompt asking for what matters most, latest developments, risks, contrarian angles and actionable takeaways.

This turned scattered searching into a concise briefing that saved me hours each day. As someone who researches every day for my job, this time saver was priceless.

Here's the prompt I used: Research this [inserted topic] like a senior analyst. Give me: What matters most, latest developments, contrarian angle, risks and actionable takeaway.

3. Brainstorming agent

Using ChatGPT to help me know what to actually research has been a time-saver, too. I recently noticed that ChatGPT-5.5 is much less of a people-pleaser than previous models. The agent actually told me an idea I had wasn't a good one, and it made me happy to see the AI finally being honest.

Knowing that I can lean on an agent to actually provide me with honest feedback, is something I've been waiting for with OpenAI's models. The key here is using the AI agent to help build momentum with solid ideas.

From there, I can vibe code my ideas and bring them to life in hours rather than days or weeks.

Here's the prompt I used: You are my elite brainstorming partner. Your job is to help me generate sharp, original, high-upside ideas quickly. Topic: [X], Goal: [X], Audience: [X], Tone: [X]. Give me: 15 strong ideas ranked from safest to boldest,2. 5 unexpected contrarian angles, 3 ideas most likely to succeed and why, one idea that feels risky but could be a breakout hit, be specific. Push for originality and usefulness.

4. Calendar optimizer

I pasted my schedule and asked ChatGPT to reorganize it around energy levels, focus blocks and wasted transitions. Once again, although Gemini has Calendar integration, I discovered that I prefer ChatGPT because it just "knows" me better thanks to its memory.

Using this principle, the AI agent suggested grouping shallow tasks together and protecting specific windows for deep work. I also discovered I can upload several calendars at once including my family's calendar and the school calendars (for two schools) and it added them seamlessly.

Here's the prompt I used: Act as a performance coach. Based on this schedule, redesign my week for maximum focus, less stress and fewer wasted hours. Schedule: [pasted calendar].

5. Decision agent

For small decisions that can quietly eat time, I used prompts comparing options based on upside, downside, hidden costs and long-term impact.

This helped reduce overthinking, which is something I'm known to do.

Here's the prompt I used: Help me decide between Option A and Option B. Evaluate: Short-term upside, long-term upside, hidden risks, opportunity cost. Best choice if I were thinking clearly

What I noticed after just a couple days

No, I didn’t suddenly work from a beach while robots ran my life. But I did notice three real changes. Although many tasks aren't hard (cleaning out my inbox, for one), they’re just annoying. Email sorting and organizing schedules are time vampires that drain energy. AI removed much of that drag.

I also noticed better focus thanks to getting the repetitive work out of the way. I had more attention left for judgment, creativity and higher-level decisions. This ultimately increased momentum and workflow day after day.

Across the week, I estimate these systems saved me roughly 7-10 hours of work that normally takes so much time I have to do it outside of work hours. Faster research, reduced task switching, fewer stalled decisions and quicker email handling gave me time back that I didn't even know was being wasted.

As a mom who works full time, even saving one hour per day adds up quickly. But I need to be frank about something important: AI still needs supervision. Sometimes the "brainstorming" was too generic and the recommendations lacked context. The agent also stumbled a bit with deciphering between "urgent" and "needs reply." Plus, there are just certain tasks that simply require human taste, nuance or judgement and I had to step in. That’s why I think the best use of AI isn’t replacement. It’s leverage.

Bottom line

If you're ready to try this for yourself, start by picking one recurring pain point this week. Maybe it’s inbox overload, messy scheduling or slow starts on writing projects. Then build one simple AI workflow around it.

Naval Ravikant’s leverage rule feels more relevant than ever. We’re entering an era where one person with the right systems can accomplish far more than someone working manually. If you try these prompts, let me know in the comments. And stay tuned — I’ll be sharing more advanced prompts soon to help you run your business and automate even more workflows.

Follow Tom's Guide on Google News and add us as a preferred source to get our up-to-date news, analysis, and reviews in your feeds.

More from Tom's Guide

• AI voice cloning is everywhere — here's why Taylor Swift’s new ‘Legal Shield’ is a blueprint for your digital safety
• I used ChatGPT to apply Warren Buffett’s ‘two-list’ rule — and it changed how I set goals
• I asked ChatGPT to use Elon Musk’s ‘Relevance Rule’ to fix my memory — and I’m never going back to notes