Ingest Learnings of a Lifetime into Your Work in Seconds

Tap into the internet’s collective knowledge and insert it into your tasks.

When I sat down to update the About page for We Eat Robots, I wasn’t sure what to improve. Then it hit me: Substack and other folks must have outlined tips for exactly this task. What if I take those best practices, attach them to a prompt, and let AI do the hard updating work, taking into consideration guidelines from the best minds on this topic?

The internet has how-tos and tips for every task, job, or challenge imaginable. AI tools allow you to synthesize and ingest this knowledge into your work.

This approach works for anything—creating a lesson plan, preparing for a job interview, making a workout plan—and takes just three steps. I'll walk you through them now, using my We Eat Robots Substack About page as an example.

Step 1: Research best practices using Google or Perplexity

Google “best practices Substack About page,” and you get tons of results. You can select the sources you trust (for example, Substack’s own recommendations) and curate the tips you like into a Word document (which we’ll attach to our prompt in the next step).

A more powerful option is to use Perplexity. Perplexity is like a cross between ChatGPT and Google. It optimizes your search query with AI, goes out to find all relevant search results, and then synthesizes them into one answer (with source attribution!). ¹

Both approaches work, but the Perplexity one is much, much faster.

Perplexity’s synthesized answer from 16 sources on best practices and tips for writing a Substack About page.

Step 2: Attach the curated best practices document to your AI prompt

With my best practices document ready, I attach it to my prompt.² Your input will vary from mine depending on the task you're working on; mine looked like the one below since I already had an (outdated) version of my About page.

Please review the current draft of my Substack About page for We Eat Robots (which I've attached). Compare it against the attached best practices and tips for writing effective "About" pages for Substack publications. 

Let me know what needs improvement and what's missing, and propose how to address these issues before proceeding with writing a new draft of the About page.

With this prompt, the AI will give me a plan of action, which I can then comment on (if necessary) before giving the green light to move ahead (in my case, writing a new draft).

You can use the prompt template below as a starting point for your own work:

I'm working on [task/project description]. To ensure I follow industry best practices and expert advice, I've attached a document containing curated tips and guidelines for this type of work.

Please review my current [draft/outline/plan] (attached) and provide feedback on the following:

1. How well does it align with the attached best practices? 
2. What specific areas need improvement to better adhere to these guidelines? 
3. What key elements or sections are missing that are recommended in the best practices? 

After reviewing the feedback, please generate an improved version of my [draft/outline/plan] that incorporates the attached best practices while maintaining my unique voice and perspective.

💡 You can also run the template above in two steps. First, get the feedback, then run the part from “After reviewing the feedback…”. This approach allows you to absorb and discuss the AI’s suggestions before telling it to generate an improved version.

Step 3: Refine, double-check, and push the AI

Once you receive your output from the previous step, don't hesitate to go back and forth with your AI tool. Like a human coworker, you can't expect to get everything perfect on the first try.

In my case, I followed up with a little nudge to play more on the eating part of the We Eat Robots name:

That's pretty good, but can you work in a few nods again to the eating theme?

The AI happily obeys your every wish, so don't hesitate to speak your mind and refine the output as often as you like.

⚠️ Pay attention to details. AI tools can still make mistakes, overlook things, and forget. For example, Claude, the AI model I currently prefer, sometimes overlooks or “forgets” details from earlier in a conversation. It usually knows what I mean and corrects itself when I point this out.

Don’t forget your own experience and learnings

The internet’s collective knowledge is pretty great, but don’t discount your own wisdom. You can also use this technique to attach a document with personal principles, values, or learnings for the AI to check your work against.

The beauty of this approach is that a sufficiently intelligent model like Claude can grasp whatever you attach (even if it’s long!) and then provide nuanced feedback and suggestions.

Give it a try, and don’t hesitate to hit Reply if you have questions or run into issues.

👋 Till next time,
Tim

1. I wrote an article on the Animalz blog with a more in-depth explanation and tips for using Perplexity: Goodbye Hallucinations, Hello AI Research: Advanced Tactics for Content Teams. (It's focused on content marketing but should still give you some ideas on how to use Perplexity even if you're in another field.)

2. Why not use the knowledge inherent in the AI models? They've read everything that matters on the internet, so they also know about the best practices in most fields.

That's true, but it's hard to check and control which knowledge and directives they follow. By attaching your own curated list, you force the AI to focus on your input.