Almost Timely News: 🗞️ Revisiting the Basics of AI (2025-08-03)

5 key basics for you to revisit and share

Almost Timely News: 🗞️ Revisiting the Basics of AI (2025-08-03) :: View in Browser

The Big Plug

👉 My new book, Almost Timeless: 48 Foundation Principles of Generative AI is now available!

Content Authenticity Statement

100% of this week's newsletter was generated by me, the human. You will see bountiful AI outputs in the video. Learn why this kind of disclosure is a good idea and might be required for anyone doing business in any capacity with the EU in the near future.

Watch This Newsletter On YouTube 📺

Click here for the video 📺 version of this newsletter on YouTube »

Click here for an MP3 audio 🎧 only version »

What's On My Mind: Revisiting the Basics of AI

This week, let’s get back to basics for a bit. I’ve been on the road a ton, talking with actual people who have actual problems they want to try solving with AI.

The thing I’ve observed most? People are still using tools like ChatGPT as though it were 2023, instead of taking advantage of the tools as they are today. They’re frustrated. They’re confused. They’re overwhelmed.

So this week, let’s fix that. Let’s review the modern basics of generative AI, basic tips and techniques we should all be using. By the time you’re done with this issue, you’ll have the modern basics you need to get far more out of any AI tool. And if you already know the basics, then please share this with two colleagues or friends who might not.

Part 1: Mechanics of LLMs

Before we begin techniques, let’s discuss technology. One of the fundamental misunderstandings about how LLMs (large language models) - the engines that power tools like ChatGPT - work is that your prompt is the last thing you said to it.

With human conversation, you write a text to a friend, they write a text back, and so on. You go back and forth, and you think the conversation is the last couple of things that you said to your friend. That's how we work. We don't keep the entire chat history in mind.

That's not how machines work.

If you open up the hood and look inside, what you find is that the ENTIRE conversation you’re having with the AI is the prompt. Everything you’ve said in that chat becomes the next prompt, along with your latest prompt - the whole conversation gets reloaded and reprocessed every time.

Imagine you were talking to your human friend and they texted you the entirety of your conversation to that point in the message. You'd be like, "Are you okay? What in the world are you doing?"

This has profound implications for how these tools work. First, everything counts (one of the 48 principles I share in my new book, Almost Timeless: 48 Foundation Principles of Generative AI) in a conversation. So if you’ve had a chat about blackberry flaugnarde and then you switch topics to B2B marketing, all that conversation about the dessert is affecting your B2B marketing. Sometimes that might be good, sometimes that might be bad.

For best results, a chat should equal a task. Changing tasks? Start a new chat.

Because the AI also takes into account the entire conversation, the longer a conversation goes on, the more the AI has to work. After a certain point, it starts to lose precision. You’ll notice its responses degrading. When you start to notice that, it’s time to start a new chat.

Finally, there is no such thing as “forget everything we’ve discussed so far and do this instead” or similar style of directions. While you can tell an AI that, it might even confirm, "sure, I will follow those directions"... but it won't. It’s the equivalent of saying “don’t think of an elephant”. If it’s in the chat history, the AI is still considering it. If you’ve got wrong information in a chat, start a new chat.

New chats: start early, start often.

Part 2: Prompting 101

Now that we’ve got the basic mechanics out of the way, let’s move into prompting. The closest thing I have to a magic trick or silver bullet for prompting AI I believe it's principle 26 in the book. is a single sentence, which you should paste at the end of every prompt:

Ask me one question at a time until you have enough information to complete the task.

This one sentence changes the behavior of AI considerably. All LLMs are trained on three basic guidelines: be harmless, be helpful, be truthful - in that order. Harmless means don’t give the user answers to obviously harmful questions like “how do I do this Very Bad Thing™?” (Fun sidebar: uncensored models will answer those kinds of questions very fluently) Helpful means try to fulfill the user’s request. And truthful is a distant third, try to give truthful answers.

Truthful is the most difficult for LLMs because truth [a] is squishy. Truth very much depends on identity and ideology these days, for good or ill. And [b] truth and probability are not the same things, and LLMs deal in probability. But that’s a topic for another time.

Where the problem is for most users of tools like ChatGPT is that second pillar, helpful. AI tools are trained to be helpful, not thoughtful. They attempt to answer a request based on what they’re given, and when they aren’t given enough information, they try to fill in the blanks themselves, based on the knowledge and data they’ve been trained on and the tools they have access to (like web search).

What this means is that if we give it a directive, like “write a blog post about B2B influencer marketing”, it’s going to be helpful. It’s going to infer or search for data to fulfill the request and carry it out as quickly as possible, in its attempt to be helpful.

Imagine a human subordinate who was afraid to ask you a question or get clarification. They’d just nod, say “yes ma’am, right away ma’am” and try their best to fulfill your assignment - and they’d be set up to fail because the blog post prompt is a terrible directive.

By adding that sentence to the end of your prompts, you instead change it from being mindlessly helpful to being thoughtfully helpful. AI has no agency of its own. AI has less agency than a mosquito, less agency than your dog. AI will not think unless you tell it to, so by telling it to ask you one question at a time until it has enough information to do the task, you are giving it permission to ask questions and gather more information. You are giving it permission to think.

You have to give AI permission to think.

Why one question at a time? Two reasons. First, humans get overwhelmed when confronted with a page full of questions. One question at a time is less overwhelming for many of us. Second, go back to our mechanics of how LLMs work. As we have a back and forth conversation, the entire conversation is the next prompt, so we can triangulate faster on a helpful response by answering one question at a time rather than all at once.

If you took nothing else away from today’s newsletter, that one sentence will change how useful AI is to you. iI will double AI's usefulness to you easily.

But we’re not done. Let’s move on.

Part 3: Knowledge Blocks and Context Engineering

The second thing that will dramatically improve the results of AI is what I call knowledge blocks (Principle 25 from… well, you know). This has a fancy new term in the nerd herd: context engineering. I’m not a fan of that term because [a] it’s confusing and [b] it sounds elitist and more important than it is.

Knowledge blocks are exactly what they sound like: blocks of knowledge, of data, that you have on hand and that you can add to prompts, like Legos. You keep these in little text files, PDFs, or even in a notebook and then just drop them into relevant prompts like Emeril tossing garlic into the pan. BAM! Better results.

Here’s a short selection of the knowledge blocks I recommend you have written out or available:

• About you/your company: who are you? What do you do?

• How do you market: assuming you’re involved in marketing, how do you do your marketing? What channels, what methods?

• Reports: while we never want to let generative AI tools do math (they’re functionally unable to), we absolutely do want them to ingest completed reports, from Google Analytics to your CRM.

• Competitive data: who are your competitors? What are their strengths, weaknesses, opportunities, and threats?

• Customer data: who is your ideal customer profile? Who is your current customer profile? What are their needs, pain points, goals, and motivations?

• Brand style guides: does your brand have a graphic identity? What are your brand's fonts? What are your brand's logos? Writing style guide and requirements?

• Anything you copy and paste more than twice: this is the big one. If you find yourself copy pasting the same chunks of text over and over again, that’s a prime candidate to turn into a knowledge block.

You don’t need anything special to store or use knowledge blocks. There’s no software to buy beyond what you already have for managing chunks of text. I personally use a free, open source app called Joplin because it fits my workflow, but you can use Google Keep, OneNote, Evernote, OneDrive, whatever.

Suppose you don’t have these blocks? How could you get them? One of my favorite ways to do that is to take the Voice Memos app on your phone and just talk, record the information that's in your head, and then use any transcription tool to turn it into a knowledge block.

But there's a even better way, an AI tool. And that AI tool is called Deep Research.

Part 4: Deep Research

Almost every major AI software tool has a Deep Research capability - Microsoft Copilot, Google Gemini, ChatGPT, Perplexity, Anthropic Claude, you name it. They may have slightly different names, but they’re all functionally the same thing - research agents that go out and, given the instructions you’ve provided them, try to assemble a lengthy document about your subject of inquiry.

I’m incredibly enamored with Deep Research as an AI capability We covered that in uh a past newsletter., because it solves a major gap in generative AI. One of the cardinal rules of AI (principle 10 in my book) is that the more data you bring, the better AI performs. Not just in terms of better results, but reduced hallucinations and errors. If you don’t have a knowledge block - or you maybe don’t trust the ones you have - you can have Deep Research tools build them for you.

In fact, I think it’s worth using Deep Research tools even when you DO have your own existing knowledge blocks because it gives you an outsider’s perspective on the topic. For example, I would encourage you to commission a Deep Research report on yourself or your company. Here’s an example prompt, which you should pair with the Trust Insights CASINO Deep Research Prompt Framework:

Let’s build a company research profile about (your company name) found at (your company website). We want to build a report that explores the company, its value proposition, who its customers are, who its competitors are, and its position in the market. Using the Trust Insights CASINO Deep Research Framework attached to this prompt, ask me one question at a time until you have enough information to build a complete, thorough CASINO prompt.

Then attach the PDF, answer the questions, and see what comes up. If the results are great? Congratulations, you now have a Knowledge Block that is fantastic for use in other prompts. Think about what that would mean if you had a knowledge block like this. You could just drop this whole thing in with a screenshot of your Google Analytics and your AI tool would not only be able to understand the Google Analytics data, but it would understand it in the context of your company.

If the results are poor? Now you have an idea of how much work you need to do optimizing your presence in AI tools. (See this back issue for more about how to do that)

Here’s the thing to remember about Deep Research tools: under the hood, they’re research agents. They can perform almost any kind of research-type task where you need someone to go out, search a bunch of stuff, and come back with an synthesis of it: customer profiles, competitive reports, strategy analyses, public data collection, you name it.

My favorite use case is “manual on demand”. When I get stuck trying to solve a problem and I can’t make headway, I commission a Deep Research report that gives me a step-by-step manual of how to solve it. Even on the rare occasion when it’s not correct, it’s still almost always useful for getting me unstuck.

But back to where we were - Deep Research is how you can manufacture knowledge blocks when you don’t have them. Their usefulness is directly proportional to how thorough your prompt is, so be sure to use the CASINO framework or a similar Deep Research framework to maximize your chances of a useful output.

Suppose you get a great result that needs a few tweaks or needs some augmentation. Do you need to start over?

If your AI tools has a Canvas, you do not. Let’s take a look.

Part 5: Canvas Tools

Most AI tools these days have some kind of workspace. ChatGPT and Gemini call it a Canvas. Claude calls it an artifact. I’m not sure what other systems call theirs, but they’re all functionally the same thing - a workspace where you can collaboratively edit the output of AI, almost like a miniature word processor.

Canvas tools let you edit a generative AI output, as well as prompt the AI to revise or refine specific sections of your output (rather than generate a whole new version). For example, you might have a competitive analysis and one paragraph just misses the mark in terms of tone. You can highlight the paragraph and give the AI tool a prompt to fix it.

Most Canvas tools also have built-in controls to change readability/reading level, style, tone, etc. Those icons in the interface? Hover over them and see what they do. There’s a good chance at least one of them will be useful and you didn’t even know it was there.

Suppose you’re writing a report for an executive, and you know the executive responds to a certain style of communication, or they’re not a strong reader. You can change the reading level of an entire document just by moving the slider or selector, and the AI tools will make those changes to the document.

Even more powerfully, Canvas tools are often capable of rendering certain kinds of code, like HTML and CSS, the technologies that make up web pages. While they are not graphic design tools for the most part, they can create fairly sophisticated web page outputs that look terrific. If you have, for example, your brand’s logo URL in your style guide, along with your preferred fonts and colors, you can put that in a prompt and have AI create an output that’s branded with your style.

Here’s the prompt you would copy and paste in a chat. You’ll note that I explicitly tell it to use the Canvas, as well as turning on the feature.

Using the attached style guide, render the previous output in the Canvas using HTML, CSS, Tailwind, CDNJS. Ensure the output meets the style guide, including the logo at the URL specified in the style guide, fonts or nearest closest match in Google Fonts, and other style requirements. Show your output in the Canvas.

You can integrate this as part of your initial prompt or use it later in the conversation, once you have an output you want to save.

What if your preferred AI tools doesn’t have a feature like this? Let’s make this our final point.

Part 6: No Mo‘ FOMO

In the last TWO WEEKS, the following MAJOR AI model updates occurred:

• Google Gemini 2.5 Deep Think + Opal

• Moonshot AI Kimi K2 Instruct

• zAI GLM 4.5 family

• Alibaba Qwen 3 refresh and new Coder

• Mistral Magistral, Codestral, Devstral, and Voxstral

• LG Exaone 4

• Cohere Command A Vision

• Wan 2.2

• Flux 1 Krea

• Hunyuan 80B

Now, while many of these may not have made headlines in business publications, they are all big deals in the AI nerd herd. What this all means is this: AI is advancing so, so rapidly that if your preferred tool or ecosystem doesn’t have a useful feature today that another platform does, there’s a very good chance in 3-6 months that your tool/ecosystem will.

For example, the biggest general laggard in terms of features is Microsoft Copilot, for a bunch of reasons I won’t get into here. However, Copilot Agents (GPTs) just debuted not too long ago, and the new Copilot Research (Deep Research) is competitive with other Deep Research platforms. If you're feeling left out, don't worry. You will get new stuff fairly soon.

If we look at this chart from Artificial Analysis, what we see is pretty clear: in terms of raw capabilities and intelligence, virtually every major model is comparable.

Nerds like me will go around chasing every shiny new object because it’s fun and exciting. If chasing shiny objects is not at all fun for you and is instead exhausting or overwhelming, take a break and recognize that the ecosystem you’re using will catch up quickly for any useful feature.

Overall, AI’s capabilities, in terms of the complexity of tasks it can handle, doubles roughly every 6 months. A task AI couldn’t do a year ago, it can probably do in some capacity today if the model’s architecture permits it. So put the FOMO aside if you’re not an AI nerd and focus on getting the most out of the tools you already have. I promise you that if it’s a major player (ChatGPT, Gemini, Claude, Copilot, DeepSeek, Qwen, Mistral), it’s smart enough.

Part 7: Wrapping Up

The purpose of this issue of the newsletter was to provide the barebones basics for modern use of these tools, from prompting 101 to knowledge blocks to modern utilities like the Canvas. Again, you may already have had all these basics down if you’ve been a longtime subscriber, so instead please share this with someone who doesn’t have the basics down.

Where do you go from here? Go do stuff. Try things out. Ask questions of the AI tool you're using. Break things. But recognize most of all that the basics of using AI well are all about having the tools prompt you.

I can't emphasize this enough. It is not just about you prompting AI. If you want to get great at AI, have AI prompt you, for more, better thinking, more, better data, and more, better ideas.

If you'd prefer more structure, here are some commercial resources:

• My book

• Mastering Prompt Engineering for Marketers Course

• Generative AI Use Cases for Marketers Course

How Was This Issue?

Rate this week's newsletter issue with a single click/tap. Your feedback over time helps me figure out what content to create for you.

• 😄 Good

• 😐 Neutral

• 😦 Bad

Here's The Unsubscribe

It took me a while to find a convenient way to link it up, but here's how to get to the unsubscribe.

If you don't see anything, here's the text link to copy and paste:

https://almosttimely.substack.com/action/disable_email

Share With a Friend or Colleague

If you enjoy this newsletter and want to share it with a friend/colleague, please do. Send this URL to your friend/colleague:

https://www.christopherspenn.com/newsletter

For enrolled subscribers on Substack, there are referral rewards if you refer 100, 200, or 300 other readers. Visit the Leaderboard here.

Advertisement: Bring Me In To Speak At Your Event

Elevate your next conference or corporate retreat with a customized keynote on the practical applications of AI. I deliver fresh insights tailored to your audience's industry and challenges, equipping your attendees with actionable resources and real-world knowledge to navigate the evolving AI landscape.

👉 If this sounds good to you, click/tap here to grab 15 minutes with the team to talk over your event's specific needs.

If you'd like to see more, here are:

• My speaker preview reel (YouTube)

• A full-length keynote you can enjoy

ICYMI: In Case You Missed It

This week, Katie and I ranted about vibe coding and what non-coders are doing wrong.

• In-Ear Insights: Everything Wrong with Vibe Coding and How to Fix It

• How to Use AI to Solve Math Problems Without the Risk of Errors

• How AI Reads Your Brand: The Mind-Blowing Word Web Behind Every Response

• How to Spot Real AI Experts (and Avoid the ChatGPT Posers)

• Why Google’s Ironwood TPU Could Break NVIDIA’s AI Chip Monopoly

• Unlock the Power of AI Without Coding: Google Opal’s No-Code Revolution

• Almost Timely News: 🗞️ The Next Generation of AI Agents (2025-07-27)

• Is SEO Worth Doing? Rode CallMe, and Google Opal for AI Orchestration

• So What? Summer Makeover Series: Using Generative AI to revamp case studies

• INBOX INSIGHTS, July 30, 2025: AI Integration, Google Tag Gateway

Skill Up With Classes

These are just a few of the classes I have available over at the Trust Insights website that you can take.

Premium

• 💥 New! Generative AI Use Cases for Marketers

• 👉 Mastering Prompt Engineering for Marketers

• 🦾 Generative AI for Marketers

• 👉 Google Analytics 4 for Marketers

• 👉 Google Search Console for Marketers (🚨 just updated with AI SEO stuff! 🚨)

Free

• New! Generative AI for Tourism and Destination Marketing

• Powering Up Your LinkedIn Profile (For Job Hunters) 2023 Edition

• Building the Data-Driven, AI-Powered Customer Journey for Retail and Ecommerce, 2024 Edition

• The Marketing Singularity: How Generative AI Means the End of Marketing As We Knew It

Advertisement: New AI Book!

In Almost Timeless, generative AI expert Christopher Penn provides the definitive playbook. Drawing on 18 months of in-the-trenches work and insights from thousands of real-world questions, Penn distills the noise into 48 foundational principles—durable mental models that give you a more permanent, strategic understanding of this transformative technology.

In this book, you will learn to:

• Master the Machine: Finally understand why AI acts like a "brilliant but forgetful intern" and turn its quirks into your greatest strength.

• Deploy the Playbook: Move from theory to practice with frameworks for driving real, measurable business value with AI.

• Secure Your Human Advantage: Discover why your creativity, judgment, and ethics are more valuable than ever—and how to leverage them to win.

Stop feeling overwhelmed. Start leading with confidence. By the time you finish Almost Timeless, you won’t just know what to do; you will understand why you are doing it. And in an age of constant change, that understanding is the only real competitive advantage.

👉 Order your copy of Almost Timeless: 48 Foundation Principles of Generative AI today!

Get Back to Work

Folks who post jobs in the free Analytics for Marketers Slack community may have those jobs shared here, too. If you're looking for work, check out these recent open positions, and check out the Slack group for the comprehensive list.

• Data Analyst (Crm) at NMQ Digital

• Data Analyst (Sql/Data Mining) at Kelly Science, Engineering, Technology & Telecom

• Digital Marketing Strategist at Thomson Reuters

• E-Commerce Associate/Analyst at Riverside Insights

• Marketing Analyst at Robert Half

• Marketing Analytics Manager at Twist Bioscience

• Sales And Marketing Specialist at Gibbons Group

• Senior Data Analyst at Coalition, Inc.

• Senior Data Analyst, Growth Marketing Analytics at Omada Health

• Senior Marketing Analyst, Member at Carrot

• Senior Product Manager Prm, Reporting & Insights at Muck Rack

• Senior Strategy Associate at Angi

Advertisement: Free AI Strategy Kit

Grab the Trust Insights AI-Ready Marketing Strategy Kit! It's the culmination of almost a decade of experience deploying AI (yes, classical AI pre-ChatGPT is still AI), and the lessons we've earned and learned along the way.

In the kit, you'll find:

• TRIPS AI Use Case Identifier

• AI Marketing Goal Alignment Worksheet

• AI Readiness Self-Assessment (5P & 6Cs)

• 12-Month AI Marketing Roadmap Template

• Basic AI ROI Projection Calculator

• AI Initiative Performance Tracker

If you want to earn a black belt, the first step is mastering the basics as a white belt, and that's what this kit is. Get your house in order, master the basics of preparing for AI, and you'll be better positioned than 99% of the folks chasing buzzwords.

👉 Grab your kit for free at TrustInsights.ai/aikit today.

How to Stay in Touch

Let's make sure we're connected in the places it suits you best. Here's where you can find different content:

• My blog - daily videos, blog posts, and podcast episodes

• My YouTube channel - daily videos, conference talks, and all things video

• My company, Trust Insights - marketing analytics help

• My podcast, Marketing over Coffee - weekly episodes of what's worth noting in marketing

• My second podcast, In-Ear Insights - the Trust Insights weekly podcast focused on data and analytics

• On Bluesky - random personal stuff and chaos

• On LinkedIn - daily videos and news

• On Instagram - personal photos and travels

• My free Slack discussion forum, Analytics for Marketers - open conversations about marketing and analytics

Listen to my theme song as a new single:

• Apple Music

• Spotify

• YouTube

Advertisement: Ukraine 🇺🇦 Humanitarian Fund

The war to free Ukraine continues. If you'd like to support humanitarian efforts in Ukraine, the Ukrainian government has set up a special portal, United24, to help make contributing easy. The effort to free Ukraine from Russia's illegal invasion needs your ongoing support.

👉 Donate today to the Ukraine Humanitarian Relief Fund »

Events I'll Be At

Here are the public events where I'm speaking and attending. Say hi if you're at an event also:

• AMA Pennsylvania, York, August 2025

• SMPS, Denver, October 2025

• Marketing AI Conference, Cleveland, October 2025

• MarketingProfs B2B Forum, Boston, November 2025

There are also private events that aren't open to the public.

If you're an event organizer, let me help your event shine. Visit my speaking page for more details.

Can't be at an event? Stop by my private Slack group instead, Analytics for Marketers.

Required Disclosures

Events with links have purchased sponsorships in this newsletter and as a result, I receive direct financial compensation for promoting them.

Advertisements in this newsletter have paid to be promoted, and as a result, I receive direct financial compensation for promoting them.

My company, Trust Insights, maintains business partnerships with companies including, but not limited to, IBM, Cisco Systems, Amazon, Talkwalker, MarketingProfs, MarketMuse, Agorapulse, Hubspot, Informa, Demandbase, The Marketing AI Institute, and others. While links shared from partners are not explicit endorsements, nor do they directly financially benefit Trust Insights, a commercial relationship exists for which Trust Insights may receive indirect financial benefit, and thus I may receive indirect financial benefit from them as well.

Thank You

Thanks for subscribing and reading this far. I appreciate it. As always, thank you for your support, your attention, and your kindness.

See you next week,

Christopher S. Penn