From Data -> Knowledge

Tips for getting the most out of your knowledge with AI; steps to set up a feedback-giving bot

Welcome to another edition of the best damn newsletter in AI.

This free newsletter is designed to keep you ahead of the curve and open your mind to using AI in your work and business.

Digging deep in AI for work or AI Operations? Take a look at our membership.

Our #1 goal is to be useful. So please shoot us an email šŸ“© if you have questions or feedback, and especially if you implement something we share!

Here's what we're covering today:

  • Tips for taking full advantage of your knowledge

  • Using AI to give you feedback

  • Elon Muskā€™s progress and OpenAIā€™s saga aftermath

... and if someone forwarded this email to you, thank them šŸ˜‰, and subscribe here!

Letā€™s get to it! šŸ‘‡

TODAY'S PERSPECTIVE

Taking Full Advantage of Your Knowledge

Today we're talking about one of the most valuable assets in AI today.

Knowledge.

You might be thinking, "What knowledge?" Well, you'd be surprised by the depth of knowledge that already exists within your company.

And with generative AI and the right knowledge management strategy - you can unlock essentially perfect context and perfect memory for your work.

Crazy sounding, right?

Well, it's actually becoming more attainable than you'd think.

As more teams are experimenting with knowledge management via Custom GPTs, here's our quick tips:

1. Tap your existing knowledge sources

Businesses we work with are often surprised by how much knowledge they have already at their finger tips. Oftentimes, your first order of business should be extracting as much value as you can out of what you have, like:

  1. Use all of your past social media posts to gain context on your company's points of view and tone of voice.

  2. Use sales transcripts and customer support email threads to get full context on your customers and product

2. You can create new knowledge

One of the best methods to creating new "knowledge" is actually just better documenting knowledge you have. For example, answering commonly asked questions and creating a knowledge base of the answers.

Also remember, that every task often creates more knowledge.

For example, each time you have a conversation with ChatGPT, you're creating new knowledge about your task, project, or business. Capture that and use it.

3. Use your knowledge + data together

Structured data that lives in databases and excel files is only becoming more valuable. Many teams are already connecting AI directly to an outside database. But with knowledge, you can take it one step further. A great recent benchmarking study from data.world showed that adding in knowledge (via a graph database or other context mechanisms) are proving to up to 3x the performance of using an LLM with normal database tables.

4. Plan for longer memory

Now, you might be worried about AIā€™s short term memory. But companies like OpenAI and Anthropic are working hard to increase the context limits, or the amount of information the AI can remember at one time. Anthropic just unveiled a 200K context window, after OpenAI unveiled 128K a few weeks ago.

5. Remember that AI treats ā€œknowledgeā€ in a unique way

Long documents where you want AI to have all of the information might not be the best approach. Summaries might be better, as AI has to choose which parts to pay attention to when it generates your response.

USE CASE

Roast yourself (nicely!)

As our friend Nicole said, feedback comes a little nicer when its coming from an AI friend. We couldnā€™t agree more.

Thatā€™s why one of our favorite use cases is using AI to assess, rate, or give feedback on work - before it goes out to our team, to customers, etc.

Hereā€™s the steps to build your own Feedback Giver Prompt.

  1. Tell ChatGPT its job is to review videos and evaluate the quality of that video

  2. Tell it that it can give a rating option from Weak, OK or Strong

  3. Give it criteria for what makes a Strong video, versus an OK or Weak video

  4. Add in examples for good measure

With some iterating, you can have a prompt that gives you a rating and if you ask for it - the feedback on why it gave that rating.

And if you want Rachelā€™s take on the experience of making a feedback bot that was a little too good.

HAPPENINGS

Reminder: Custom GPTs for Business Live Workshop - This Thursday!

Weā€™re hosting a workshop 100% focused on how to build your first valuable Custom GPT. Youā€™ll even get with a personalized custom GPT plan on how to set one up for a use case of youā€™re choosing.

Note: this workshop is very beginner-friendly!

LINKS

For your reading list šŸ“š

Elon Musk's AI ventures are making waves...

Weird AI use cases and good points...

The OpenAI sagaā€™s aftermath...

And if youā€™re really nerdyā€¦

  • Securing generative AI is becoming a must as organizations lean more on its capabilities. Check out why AI security platforms are now essential.

  • California's Privacy Protection Agency is drafting AI regulations, including opt-out and access rights. They're all ears for public input.

That's all!

We'll see you again on Tuesday. Thoughts, feedback and questions are much appreciated - respond here or shoot us a note at [email protected].

... and if someone forwarded this email to you, thank them šŸ˜‰, and subscribe here!

Cheers,

šŸŖ„ The AI Exchange Team