The 60/30/10 Rule for Getting More Value from AI

Hey folks, sorry for the later-than-usual post this week -- I just finished my talk at INBOUND 2025, which is now live:

Today, I want to go deeper on one of the key frameworks I discussed: how most people are dramatically underutilizing AI's potential.

There's a myth that humans use less than 10% of their brains. That's complete nonsense.

What's actually true is that most humans use considerably less than 10% of AI's potential.

I've been watching how people interact with AI tools for the past two years, and I see the same pattern everywhere: people discover a handful of prompts that work, then use variations of those same prompts 95% of the time.

"Summarize this article for me." "Write me a subject line for this email." "Help me brainstorm ideas for X."

Nothing wrong with using what works -- but that’s leaving massive value on the table.

So in this post, I want to share a simple framework that I use to approach AI:

• The 60/30/10 rule and how it works

• Practical ways to implement it starting today

• Developing AI intuition when approaching problems

—@dharmesh

How the 60/30/10 Rule Works

The framework is straightforward: divide your AI usage into three distinct buckets, each serving a different purpose in your system.

60% - Repetition

This is your bread and butter -- the AI use cases you know work reliably. These are the prompts and workflows you've tested, refined, and can count on to deliver consistent results.

Examples might include:

• Summarizing meeting notes or articles

• Writing first drafts of emails or social posts

• Generating bullet points from longer documents

• Creating basic outlines for presentations

30% - Iteration

This is where most people miss value. Iteration means taking your proven use cases and systematically making them better. You're not changing what you're asking AI to do, you're refining how you ask to get better results.

Instead of "summarize this article," you might iterate to:

• "Summarize this article for a technical audience, focusing on implementation challenges"

• "Summarize this article as a decision-maker who needs to understand ROI implications"

• "Summarize this article and highlight three specific action items I can implement this week"

10% - Experimentation

This is pure exploration -- using AI for completely new applications you're not sure will work. These are your "what if" moments where you push the boundaries of what's possible.

The key about experimentation is that most attempts will fail or produce mediocre results. That's the point. You're developing the muscle of knowing when to use AI and when not to -- wisdom that only comes from pushing these tools to their limits.

Pro tip: If an experiment doesn't work today, set a calendar reminder to try again in 3-6 months. AI capabilities are evolving rapidly, and yesterday's failed experiment might work the next day.

How to Put This Into Practice

Implementation is simpler than you might think, but it requires a bit of structure to avoid falling back into old habits.

Week 1: Reality Check

Before changing anything, spend one week tracking exactly how you currently use AI. Keep a simple running note -- I use a basic text file -- logging each interaction:

• What you asked for

• Which category it falls into (spoiler: probably 95% repetition)

• Whether the result was useful or just "okay"

Most people are genuinely shocked by how repetitive and basic their AI usage actually is. This awareness alone will change how you approach it going forward.

Week 2+: The 60/30/10 Implementation

Here's the practical math: If you typically have 10 AI interactions per day, aim for:

• 6 reliable, proven use cases (your daily workflow savers)

• 3 iterations on existing workflows (making good prompts great)

• 1 pure experimentation (trying something completely new)

Don't stress about hitting exact numbers, if you're landing somewhere in that ballpark, you're on track.

Daily Experimentation Habit

Here's an important practice: Every time you sit down to do any task, before opening the usual applications, type it into ChatGPT first and see what happens.

You should expect that most of these experiments fail or produce mediocre results. But the ones that work become part of my regular toolkit.

Keep an Experiment Log

Simply note:

• What I tried

• Whether it worked

• How I might improve it next time

This prevents you from repeating failed experiments and helps me build on successful ones.

Developing AI Intuition

After you've consistently practiced repetition, iteration, and experimentation for a few weeks, something interesting happens: you start developing "AI intuition."

What AI intuition looks like:

• "This writing task is perfect for AI because it's structured and I can provide clear examples"

• "This strategic decision needs human judgment, but AI could help me think through the options"

• "This analysis involves too much nuanced context that AI won't understand, so I'll handle it myself"

More importantly, you'll stop seeing AI as a separate tool you occasionally use and start seeing it as an integrated part of how you think and work.

That's when you automatically start considering whether AI should be part of your approach before you even start a task.

Instead of "I have a problem, maybe AI can help," you think: "I have work to do, how should AI be involved from the beginning?"

The truth is, developing this intuition -- knowing when to use AI and when not to -- only comes from pushing these tools to their limits through systematic experimentation.

If you follow the 60/30/10 framework consistently, you'll develop judgment that puts you in a completely different league.

Remember: the goal isn't to use AI more -- it's to use AI better. And "better" comes from the deliberate practice of repetition, iteration, and experimentation.

Happy building :)

—Dharmesh (@dharmesh)