If you’ve subscribed to multiple AI tools but barely use them, you’re not alone. The 10-20-70 rule shows why in this article
You pay for ChatGPT Plus.
Maybe Claude Pro too.
You read the guides. Watched the tutorials.
Got excited about AI “transforming” your work.
Three months later, you’re barely using it.
When you do use it, the results are… fine.
Not game-changing. Just fine.
Meanwhile, someone you know is 10x more productive with the exact same AI tools you have.
What’s going on?
The answer is hiding in what researchers call the 10-20-70 rule.
And once you understand it, you’ll realize why most people struggle with AI while a few people absolutely crush it.
The 10-20-70 Rule: Why Technology Alone Changes Nothing
Here’s the rule that explains everything about AI adoption:
10% of AI success comes from the technology itself – the algorithms, the models, the “AI” part
20% comes from the tech setup – the platforms, integrations, data access
70% comes from people and processes – how you actually use it, your workflows, your habits
This is data from Boston Consulting Group studying thousands of AI implementations across companies.
Now the problem with most people using AI is, they spend 100% of their energy on the 10%.
Obsessing over which AI model is “better.”
Read countless comparison articles.
You switch between ChatGPT and Claude and Perplexity and Gemini trying to find the perfect one.
Meanwhile, completely ignoring the 70% that actually determines whether AI helps you or not.
Why the 10% Doesn’t Matter (As Much As You Think)
The AI models themselves – ChatGPT, Claude, Gemini – are all incredibly powerful. The differences between them matter, but not nearly as much as you think.
Here’s what actually happens:
Let’s say you pick ChatGPT because it’s the most popular.
You open it up. You ask it a question. It gives you an answer. Cool.
Then what?
You don’t know how to integrate it into your actual workflow. You don’t have a process for when to use it vs when not to.
You haven’t built the habit of reaching for it at the right moments.
So it sits there. Unused. A $20/month subscription you barely touch.
The technology (the 10%) is working perfectly.
You’re just not doing anything with it.
The 20%: Tech Setup That Most People Skip
The 20% is your technical infrastructure.
For you, this means:
- Can you access AI from where you actually work?
- Can you upload the documents you need to analyze?
- Does it integrate with your other tools?
- Can you switch between models when you need different capabilities?
Most people don’t think about this. They sign up for one AI tool and expect it to magically fit into their work.
Then they hit friction:
Your AI can’t access the web, so you’re manually copying and pasting research.
You need to analyze a 50-page PDF but your AI has file upload limits.
You’re switching between three different AI subscriptions because each one does something the others don’t.
The technology works. Your setup doesn’t.
The 70%: Where AI Actually Succeeds or Fails
This is the massive part that everyone ignores. The 70% is about people and processes:
- Do you actually know WHEN to use AI vs when not to?
- Have you changed your workflows to include AI at the right points?
- Did you build the habit of using it consistently?
- Do you know how to ask questions that get useful answers?
- Have you figured out which tasks AI actually helps with?
Research shows that 62% of business leaders are optimistic about AI, but only 42% of frontline employees are.
Why?
Because leaders talk about AI’s potential (the 10%)
while employees deal with the reality of trying to use it in their actual work (the 70%).
What the 70% Actually Looks Like
Here how you know if you’re going to be successful with AI or not.
Failed AI Usage (Ignoring the 70%):
- You open ChatGPT when you remember it exists
- You ask vague questions and get generic answers
- You don’t integrate the outputs into your actual work
- You give up after a few mediocre results
Successful AI Usage (Nailing the 70%):
- You’ve identified specific tasks where AI saves you real time
- You’ve built AI into your daily workflow at specific trigger points
- You know how to frame questions to get useful outputs
- You’ve developed the habit of refining AI responses instead of using them raw
The difference isn’t the technology. It’s the process around the technology.
Why Most AI Implementations Fail
BCG research shows that 79% of employees believe AI-specific training is necessary.
But only 14% of frontline employees have actually received it.
Translation: Companies buy AI tools (the 10%), maybe set up some integrations (the 20%), then wonder why nobody uses them (ignored the 70%).
For individuals, it’s the same pattern:
You subscribe to AI tools ✓
You maybe watch a tutorial ✓
You never actually build it into your workflow ✗
Then you wonder why AI “doesn’t work for you” when really, you never did the 70% that makes it work.
How to Actually Make AI Work: Focus on the 70%
Stop obsessing over which AI model is 2% better. Start focusing on the 70%:
- Identify specific, repeating tasks where AI saves time
Don’t use AI for “everything.”
Pick 3-5 specific tasks you do regularly where AI actually helps.
Research. First drafts. Data analysis. Whatever makes sense for YOUR work.
- Build AI into your workflow at trigger points
Not “I should use AI more.”
Instead: “Every time I need to research a prospect, I upload their materials to AI first.”
Specific actions get specific processes.
- Develop the skill of working WITH AI outputs
AI doesn’t give you finished work. It gives you rough drafts.
Learn to treat it as an assistant and not a replacement tool.
- Track what actually saves you time vs what wastes it
Not all AI usage is useful.
Some tasks actually take longer with AI.
Figure out what works for you and your team.
Double down on what works. Cut what doesn’t.
- Make it as frictionless as possible
If you have to jump through hoops to use AI, you won’t use it.
Get your setup right (the 20%) so using AI is easier than not using it.
The OpenCraft Approach to the 70% Problem
Here’s where most AI platforms fail you:
They give you powerful technology (the 10%) but leave you to figure out the 70% on your own.
OpenCraft is built differently:
Multi-model access means you’re not constantly switching platforms and relearning interfaces. One workspace. Multiple AI models. Less friction in your actual workflow.
Document-first design means uploading and working with your actual work materials is built-in. The 20% is handled.
Workspace memory means the AI remembers your projects and context. You’re not starting from scratch every conversation. The 70% gets easier over time.
We’re giving you AI that fits into how you actually work.
