[0:00] Most people learn how to prompt Claude
[0:01] from random tips on the internet. But
[0:03] Anthropic, the company that actually
[0:05] built Claude, released their official
[0:06] prompting guide. Almost no one's read
[0:08] it. I did. And here are the seven most
[0:11] important rules straight from the people
[0:13] who created the model itself. Let's get
[0:15] into it. So, as I mentioned, these tips
[0:17] came directly from Anthropic, the
[0:19] company that made Claude. And this here
[0:20] is the blog post that I'm referring to.
[0:22] So, it walks through a series of tips,
[0:23] not just seven, many more. But I wanted
[0:25] to call out the seven that I feel like
[0:26] are applicable to the broadest audience
[0:28] where people can get direct impact of
[0:30] using these tips. And we'll start with
[0:31] tip one, which is to be clear. Now, this
[0:35] is more critical than ever for all the
[0:37] models that are relevant today, such as
[0:38] Gemini 3, GBT 5.2, Claude Opus 4.5. All
[0:42] of these models are very good at
[0:44] following instructions. So, when you're
[0:46] clear with the model, you're going to
[0:47] get a higher quality response, and the
[0:49] AI is going to work in the direction
[0:50] that you care about. Vague prompts tend
[0:52] to lead to common patterns that are kind
[0:55] of considered AI slop today. So it's
[0:57] generic outputs that aren't very useful.
[0:59] Now why does this happen? It happens
[1:02] because most models have a gravitational
[1:05] pull towards what's most common. So here
[1:08] we have gravity pulling in all the
[1:11] commonalities of the internet. And
[1:12] here's the default state. An example
[1:14] here is UIs. So often times when an AI
[1:17] creates a UI for you, it's going to have
[1:19] purple gradients and it's going to have
[1:20] rounded corners. Reason being is the AI
[1:23] was trained on millions of examples that
[1:24] had that type of aesthetic. So the AI
[1:27] thinks that if it provides you that
[1:28] aesthetic, you're going to be satisfied
[1:30] as a user if you don't provide it very
[1:32] specific instructions. And that's the
[1:34] key here. We have to be specific to
[1:37] escape the gravitational pool that is
[1:39] all those examples the AI was trained on
[1:41] historically. And that's what this tip
[1:42] highlights. So I'm going to walk you
[1:43] through a few examples of what good
[1:44] looks like. So on the left hand side we
[1:46] have a bad prompt. This is weak. So
[1:48] we're basically stating create an
[1:49] analytical dashboard. We we gave no
[1:51] specificity here. A better prompt is
[1:54] create an analytical dashboard. Include
[1:56] as many relevant features and
[1:57] interactions as possible. Go beyond the
[1:59] basics to create a fully featured
[2:00] implementation. And there's a few key
[2:02] things I want to call out here. So first
[2:03] off, we're stating I want you to include
[2:05] as many. So be as inclusive as possible
[2:07] when it comes to the features and
[2:09] interactions relevant to the dashboard
[2:10] we're trying to create. Another thing
[2:12] we're making explicit is the AI needs to
[2:14] go beyond the basics. So, it's going
[2:16] beyond what's normal and creating
[2:18] something that's fully featured that's
[2:20] relevant to the initial dashboard we're
[2:22] trying to create. And this is what a
[2:23] good prompt looks like that's more
[2:24] clear. And here's another example for
[2:26] presentations. So, here we have a week
[2:28] example that says create presentation.
[2:30] An improved version of this is to create
[2:31] a professional presentation in relation
[2:33] to our quarterly results. I want you to
[2:35] include thoughtful design elements,
[2:38] visual hierarchy, and engaging
[2:39] animations where appropriate. And as you
[2:41] can see, we're being very clear on what
[2:42] we want the AI to include in the visuals
[2:44] here, and that's key. So, this is our
[2:46] first tip, which is be clear, quick
[2:48] pause in your regular programming. This
[2:49] video is brought to you by me, as
[2:51] always. So, two quick things. First off,
[2:54] below is a 30-day AI insight series,
[2:56] completely free. You'll get 30 insights
[2:58] in your inbox how you can apply AI to
[2:59] your business and your work. The second
[3:01] thing is, if you'd like to work with me,
[3:02] there's a variety of ways you can do
[3:04] that, either through a private AI
[3:05] community or directly one-on-one. You
[3:07] can check that out below as well. Let's
[3:08] get back into the video. Our second tip
[3:10] is often something that a lot of people
[3:12] skip, which is explain why. And what do
[3:14] we mean by this? Well, when you give the
[3:16] AI some sort of context, so you ask it
[3:18] to do something. If you also add the
[3:20] intent as to why you're asking the AI to
[3:22] do this and why it's important for the
[3:24] AI to take that task on, it can actually
[3:26] infer a lot from what you've stated,
[3:28] even if you've not made it explicit. In
[3:30] other words, Claude can figure out
[3:32] things that you didn't even say as long
[3:33] as you're explicit about why you're
[3:35] asking the AI to do this in the first
[3:36] place. And that often usually deres a
[3:39] higher quality output for you. Now let's
[3:41] look at some examples. So this first
[3:43] example here is a weak example. Not a
[3:45] great prompt. We're saying write this in
[3:46] a formal tone. It can be improved. A
[3:48] better way of saying this is I want you
[3:50] to write this in a formal tone because
[3:52] remember we're emphasizing the why here.
[3:54] It's going to our board of directors and
[3:56] we need to look credible and
[3:57] professional. By providing just this
[4:00] context, the AI is going to be able to
[4:01] create something that's more
[4:02] aesthetically pleasing and aligned with
[4:04] the quality you're seeking by knowing
[4:06] this specific intent. Now, here's
[4:07] another example. So, here we have the
[4:09] weak prompt of saying, "Keep it short."
[4:10] A better version of this is stating,
[4:12] "Keep it short because, again, we're
[4:13] highlighting the Y. I'm sending this to
[4:15] my team via text message, and longer
[4:17] messages don't get read." Again, we're
[4:19] highlighting the Y, which is going to
[4:20] get better quality outputs from the AI.
[4:22] That's our second tip. And now, we'll
[4:24] move on to tip three, which is giving
[4:25] good examples. So the thing with
[4:27] examples, especially in relation to the
[4:29] state-of-the-art models today, such as
[4:30] Opus 4.5, Gemini 3, GPT5.2,
[4:34] all of these models, they follow the
[4:36] examples to the tea. So if you include
[4:37] an example in your system prompt, the AI
[4:39] is going to follow that very
[4:40] specifically. So it's important that you
[4:41] don't misdirect the AI by accidentally
[4:43] including something in your example that
[4:45] you don't want in its output. And a meta
[4:47] lesson in relation to this tip is not
[4:49] only is the example that you explicitly
[4:50] call in the prompt dictating what the AI
[4:52] gives you, but also the way that you
[4:54] write your prompt determines how the AI
[4:56] gives you an output. What does that
[4:57] mean? In the sense that if I write my
[4:59] prompt in a very playful, fun, and
[5:01] simple way, the AI is likely going to
[5:03] respond and give me an output in that
[5:04] way. If I might write my prompt in a
[5:06] very formal and structured way, the AI
[5:08] is going to do the same. So even the way
[5:10] that I write my prompt is going to
[5:12] dictate how the AI outputs its response
[5:14] to me. So it's important to note not
[5:16] just the example but also your prompt
[5:17] matter. Now on to tip four which is
[5:19] asking for the format. Not saying
[5:21] something that you don't want. You want
[5:22] to say what you do want because often
[5:24] when you look at system prompts people
[5:26] have really huge negating terms of
[5:28] saying never do this or don't do that or
[5:30] whatever else. Well with AI today since
[5:32] it follows instructions so well all you
[5:34] have to do is say what you want instead
[5:36] of what you don't want and the AI is
[5:38] likely going to do that more
[5:38] effectively. What are some examples of
[5:40] this? Well here's an example. There's
[5:42] first one is a weak example. So we say
[5:44] do not use markdown in this response.
[5:45] The reason this is weak is because we're
[5:47] using a negating term. So we're saying
[5:48] do not. A better version of this is your
[5:52] response should be composed of smoothly
[5:54] flowing pros and paragraphs. This is
[5:56] going to give us what we want instead of
[5:58] avoiding what we don't want. Another
[5:59] example is here we have a weak version
[6:01] saying make it look nice, not very uh
[6:04] not very clear. And a better version of
[6:06] this is saying use clear headers in each
[6:08] section. Bold the key takeaways. Add a
[6:10] summary at the top. We're very explicit
[6:12] about what we want. The AI is likely
[6:14] going to achieve that because it's very
[6:15] good at following instructions. So
[6:17] that's tip four. Ask for the formats
[6:18] that you want instead of saying what you
[6:20] don't want. Now on to tip five, which is
[6:22] being direct about actions. What does
[6:24] this mean? Well, often times when you
[6:25] want AI to do something for you, we need
[6:27] to avoid suggestive language. So instead
[6:30] of saying suggest, think about,
[6:32] consider, you need to be clear and
[6:34] saying I want you to change this now,
[6:35] edit this thing, or make whatever. by
[6:37] being explicit about it taking action,
[6:40] it will then take the action. But
[6:41] oftentimes the AI can default to being
[6:44] more oriented towards not taking action
[6:46] to make to ensure it's not breaking
[6:48] anything or doing anything it shouldn't.
[6:49] So if you say, "Hey, I want you to
[6:51] suggest this." It will suggest
[6:52] something. It won't change anything. If
[6:54] you want it to change something, then
[6:55] you explicitly state that you want it to
[6:56] change it. So be clear on the action
[6:59] verbs that you're using with the AI. And
[7:01] here are just some examples. So on the
[7:02] left hand side we have a weak prompt
[7:04] which we're stating can you suggest some
[7:06] changes to improve this function. The
[7:07] reason this is weak is we're stating
[7:09] suggest. We're being more careful with
[7:11] our language which means the AI will be
[7:13] careful with its actions. It's not going
[7:14] to do anything. It's only going to
[7:15] suggest what we need to do to the
[7:17] function. This is in relation to code. A
[7:19] more explicit way of doing this that's
[7:21] more actionoriented is saying change
[7:23] this function to improve its
[7:24] performance. That's when the AI goes off
[7:26] and does the action for you. Another
[7:27] example that's unrelated to code is
[7:29] proposal prompts. So a weak version of
[7:31] this is what do you think about this
[7:33] proposal? This is just looking for
[7:34] feedback, no actions. If you want an
[7:36] action to occur from the AI, we would
[7:38] say something that's more strong, which
[7:40] is edit this proposal to make the
[7:41] benefits clearer and add a call to
[7:43] action at the end. This means the AI is
[7:45] going to edit. It's going to do a thing
[7:46] because there is an action verb inside
[7:48] the prompt. So be clear about actions.
[7:50] If you wanted to suggest something or
[7:52] act acts on something, you need to be
[7:54] clear about that. Now, on to tip six,
[7:55] where Claude is actually extremely good
[7:57] at research. It's gotten much better
[7:59] over the last iteration of versions that
[8:01] that Anthropics released with Opus 4.5.
[8:03] Now, let me show you a weak and good
[8:05] example. So, a weak example here for
[8:06] research in relation to AI is saying,
[8:08] "Research my competitors." This is not
[8:10] good because it's vague and it's not
[8:11] clear on what you want. A better version
[8:13] of this is research my top three
[8:15] competitors in the home services
[8:17] industry. For each one, I want you to
[8:18] find their pricing, their main services,
[8:20] and their customer reviews. Compare them
[8:22] to my business and tell me where I have
[8:23] an advantage. So here we're explicit
[8:25] about the type of research we wanted to
[8:27] do, what we wanted to do with the
[8:28] research after it's gotten back and a
[8:30] series of other things. So this is what
[8:31] a good research prompt looks like. And
[8:33] something that's really interesting
[8:34] about the guide that was provided for
[8:36] anthropic is they actually provided a
[8:38] metaprompt. So this specific prompt
[8:39] right here can be applied to almost any
[8:41] research use case that you have and it's
[8:44] going to improve it. I've been using it
[8:45] for the last couple of days and it's
[8:46] night and day in the quality of the
[8:48] research that I'm getting getting back
[8:49] from the AI by using this prompt. So, I
[8:50] highly recommend just copying and
[8:52] pasting this from their guide and using
[8:53] it inside your research. Now, why is
[8:55] this so good? Well, there's a few things
[8:56] as to why this prompt is so useful.
[8:58] First off, we're asking the AI to
[8:59] research in a structured way. So, be
[9:01] more formal in your process. But also,
[9:03] we're asking as you gather data, I want
[9:05] you to develop competing hypotheses. So,
[9:08] multiple opinions against what's
[9:10] happening here. In addition to that,
[9:11] we're also saying, can you uh track your
[9:13] confidence levels over time and adjust
[9:15] them as you learn new things? because
[9:17] you're going to regularly self-critique
[9:19] your confidence levels as well as the
[9:21] hypotheses that you build over your
[9:22] research process. And then we're also
[9:24] noting that the research needs to be
[9:26] broken down from a complex ask to more
[9:29] manageable asks that you then
[9:30] consolidate over time. So there's a lot
[9:32] of really interesting information in
[9:34] this prompt as to why it's useful, but I
[9:35] highly recommend just using this for
[9:37] your research going forward because
[9:38] you're going to get much better
[9:39] responses back for things that truly
[9:40] matter. And then our final tip is
[9:43] Claude's really good at making documents
[9:44] specifically because it uses a claude
[9:46] skill or variety of claude skills that
[9:48] allows it to create documents that are
[9:49] more compelling. So this is going to be
[9:51] in the relation to presentations,
[9:53] animations inside of presentations
[9:54] and/or landing pages like this or any
[9:57] type of visual documents that's relating
[9:58] to a PDF, an Excel sheet, etc. So it's
[10:02] generally better at creating these types
[10:04] of documents and they tend to be
[10:05] formatted more effectively instead of
[10:07] what we used to get back in the day with
[10:08] other models. So what are some examples
[10:10] of this? Well, the first one here is a
[10:11] presentation. So a bad prompt is make me
[10:14] a presentation. A better prompt, a
[10:15] stronger one is create me a professional
[10:17] presentation on topic. You fill in the
[10:19] topic, including thoughtful design
[10:20] elements, visual hierarchy, and engaging
[10:22] animations when appropriate. So we're
[10:24] asking specifically about certain design
[10:26] elements and asking it to think hard
[10:28] about those elements. So when it
[10:29] provides us back a presentation, it's
[10:30] more compelling. So it's a better
[10:31] prompt. Another example here is a
[10:34] report. So instead of saying, "Write me
[10:36] a report." Bad prompt. A better prompt
[10:38] is create a monthly report for my team.
[10:40] Include a summary at the top, sections
[10:42] for each department, charts showing our
[10:44] progress, action items for the ne for
[10:46] the next month, and use clean formatting
[10:48] that's easy to scan. So, we're asking
[10:50] for a lot of things here. We're asking
[10:51] explicitly what should be included in
[10:53] the prompt. We're being very uh specific
[10:55] on the inclusion. We're also being very
[10:57] explicit on the formatting as well
[10:58] because we want be people to be able to
[11:00] scan this instead of having to read it
[11:02] or struggle to read through it. And
[11:03] that's our seventh tip. And as a quick
[11:05] recap, here are just seven rules that I
[11:07] pulled from Anthropic themselves on how
[11:09] you can get most the most out of their
[11:10] models. So the first one here is about
[11:13] being clear. You want to be clear with
[11:14] the model because it's very good at
[11:15] following instructions. So be clear in
[11:17] your prompt. The second thing is always
[11:19] provide your intent. Explain why as to
[11:21] why you're asking the AI to do this
[11:22] thing for you. It can infer more from
[11:24] that and give you a higher quality
[11:25] response. After that, you should give it
[11:27] high quality examples if you decide to
[11:29] give it examples because the AI is going
[11:31] to follow them to the tea. that includes
[11:33] the prompt itself of how you're writing
[11:35] it. So, be very careful on the prompt
[11:37] that you're drafting. After that, if you
[11:39] care about formatting, you want to be
[11:41] explicit about that and ask for certain
[11:42] types of formats from the AI. And that
[11:44] means we're asking the AI what to do
[11:45] instead of what not to do. After this,
[11:47] we have the importance of being direct
[11:49] with the AI. So, if we want it to take
[11:51] action, we need to be clear on it taking
[11:53] that action. If we wanted to suggest
[11:54] something, we need to be clear on that.
[11:56] Also, it's very good at research. And
[11:58] that metaprompt that I showed you, you
[11:59] should use that probably for good on all
[12:01] your research going forward with the
[12:02] specific model because it's night and
[12:03] day compared to what I've gotten in the
[12:05] past. And then finally, it's very good
[12:06] at creating documents and it's good at
[12:08] following different types of formatting
[12:10] um for generalized reports, Excel
[12:12] sheets, presentations, etc. And that's
[12:14] it. So if you enjoyed this, reshare with
[12:15] your friends. And also, as a reminder,
[12:17] two things. First off, below is a 30-day
[12:19] AI insight series, completely free.
[12:21] You'll get 30 insights in your inbox how
[12:22] you can apply AI to your business and
[12:24] your work. Second thing is if you'd like
[12:25] to work with me, there's a variety of
[12:27] ways you can do that either through a
[12:28] private AI community or directly
[12:30] one-on-one. You can check that out below
[12:31] as well. Okay, so now we know that cloud
[12:34] can create great documents. Anthropic
[12:36] even says so. But here's the thing. It
[12:38] still can't match your formatting, your
[12:40] fonts, your colors, your style. You
[12:42] still have to copy the content and fix
[12:44] it by hand in the document. I found one
[12:46] cloud feature that fixes this. Now,
[12:48] every document that comes back to me is
[12:50] exactly how I want it. Let me show you
[12:52] right here. Go ahead, click that video
[12:54] right there.