Transcripts
1. Welcome and Class Overview: Welcome to Claude for YouTube. Build a complete video brief. Now, I've been creating
briefs for YouTube, whether I'm outsourcing it to Tutorial creator for
my YouTube channel. I do that from time
to time depending on the specific software and
how much time I have, or if I'm just creating the
entire brief for myself, so I have an overview of
how to actually record the video before I actually
get started, right? Because if you start like I'm talking now and you end
up just winging it, then you go off on tangents, and that really hurts
your attention. But if you have a plan, if you have an outline,
if you have a brief, you can kind of
stick to the script, and it sounds more cohesive and to the point and together. Now, I'm Brian Guerra,
and in this class, I'm going to show you how
to use Claude to turn a rough YouTube idea into a
clear, practical creative pf. I have extensive
experience on YouTube. I've been on YouTube with four or five channels ultimately now. My main channel is around
100,000 subscribers, but I've turned YouTube into a business in multiple
different capacities, and I've been on it
for, like I said, four or five years now over
four or five channels. I've seen pretty
much every twist and turn that's come up over time, and I know not just
what to do right now, but how to stand the test
of time on the platform. Now, this is not a
class about pressing one button and asking AI to create an entire
video for you, okay? The goal here is to
combine your knowledge and your experience
because that's what separates you from the
rest of the pack, right? That's your diversification with a structured process that helps Claude give you a
more useful outline, brief, and ultimately
results, okay? So we'll build the brief
around five parts, the viewer, the promise, the angle, the packaging, and then
ultimately the delivery, right? I'll show you how to give Claude the right
source material. Make it ask useful questions before it starts actually
writing anything for you, develop accurate title
and thumbnail directions that can actually succeed, create a section by
section outline, which is crucial for
anybody recording a video, and identify claims that
obviously still need to be verified because we want
everything to be accurate. Now, this class is
designed for beginners and you can follow along
with a free clod account. You don't even need
to upgrade for this. For your class project,
you will create a complete one page brief for
AUT video of your choice, and by the end, you'll
have a clear plan that you can use to
begin scripting, filming, and then
editing your video. So let's start by defining who the video is for and what
it needs to deliver.
2. Define Your Viewer and Angle: The quality of Claude's
output depends heavily on the quality of the thinking and the prompts that
we actually give it. In this lesson, we'll begin with our own notes and experience, and then we'll make Claude ask questions before it
generates anything. That will help us
to find the viewer, the video promise, and the specific angle that we
ultimately want to take. This is the creator
brief template that we'll complete
during the class. It's got five sections, viewer promise, angle
packaging and delivery. Now, the first three
define the strategy. Packaging covers the title
and the thumbnail direction, and delivery covers
the hook examples, structure, and
conclusion, right? Now, we're going to take
this task into Claude to invent all of this
from absolutely nothing. We'll begin with a
source material that reflects our own
experience here. So for this demonstration, the video idea is how to turn one long video into three
useful short clips. Now, my notes identify
the audience, explain the method I use, include mistakes that I've seen, and define what the viewer
should be able to create. Your notes do not need
to be paused, right? They just need to
contain real information that Claude can
actually work with. I'll open a new
Claude chat and paste the source notes with
the first prompt. Notice that I'm not going to ask Claude to generate
the brief just yet. Instead, I'm going to ask
it to analyze the notes and ask up to five diagnostic
questions about the viewer, the problem, the promise, the angle, and anything
that it might be missing. Now, I'm also telling
Claude not to invent the facts or
the experience, right? Now, review the questions and answer them for your
actual knowledge. If a question reveals
a gap, don't guess. Market for research
or narrow the scope. In this example, the viewer is a creator who has
already recorded a long talking head video but doesn't know how to choose
useful short clips. The promise is not
guaranteed growth, right? It's a repeatable method
and three edited clips. The angle is to
select clips based on whether each one
can stand alone, not simply because the tool labels it as a highlight, okay? Now, I'll ask
Claude to summarize the viewer promise
and angle using only the notes and
answers and to list any remaining
assumptions separately. The last instruction makes uncertainty visible instead
of letting it disappear into, you know, a polished copy. Transfer the strongest
version into the template and
edit the language, so it sounds like something
that I would actually say.
3. Build a Creative Brief: Now that the idea is clear, we need a prompt that
gives Claude a job. We need it to use
useful contacts, and we need specific constraints and an exact output format. So in this lesson, I'll
show you the structure of the prompt and explain why each part specifically matters. This is the piece that
you'll be able to reuse for future video ideas. In this lesson, we're going
to turn raw decisions from the previous step into a reusable creative
brief prompt for Claude. The goal is not to ask Claude
to invent a video for us. The goal is to give
Claude a clear job, reliable source material, and boundaries that keep the result accurate to our own experience. I'm starting inside Claude, the same conversation where
we already define the viewer, the promise, and the angle. Keeping this work in one
conversation matters because Claude can use the
confirmed information that's already here. Before writing the
prompt, I quickly will review those
three decisions. And the viewer is a
solo YouTube creator with beginner or intermediate
editing ability. Now, the promise is a repeatable
process for selecting three self contained moments from a longer tutorial overall. Now, the angle is that
a useful short clip must work as a complete idea, not merely as just like
a specific moment. Now, I'm going to build the
prompt in four label parts. Context, task,
constraints, and output. The structure makes the
request easier to inspect and undoubtedly we use it again for other clips in the future. The context tells Claude which information is
allowed to be used. I write use confirm viewer, promise, angle, source
notes, and experience. Then I add in
important boundary. So preserve my perspective and do not invent personal stories, results, statistics,
or expertise. That sentence obviously protects the brief from sounding more authoritative or successful than obviously the source
material supports. Task gives Claude one
clear assignment. So in this case, the
task is going to develop practical options for packaging
and delivering the video. And I'm not going to ask for
vague brainstorming here. I'm going to ask for
usable decisions that can guide the
actual production. Now, next is the constraints, so I'll tell Claude to keep every idea accurate to
the source material. If I obviously don't use
promises about views, revenue, subscribers,
or virality, then that's obviously
a good thing, right? You don't want to make
any of those promises that you can obviously support. I'll also make sure that
we don't use fake urgency, fake metrics, or
unsupported claims. And then I'll ask for
meaningfully different options instead of just minor
rewrites, okay? Finally, I'll tell
Claude to flag anything that still
requires verification, and these constraints are
doing real work, right? Because without them, an
AI tool might produce impressive sounding titles
that don't fit the lesson, or it might even turn a practical tutorial into
a performance promise. Alright, we want
to use creativity, but we want to inside
truthful boundaries. Now, the output section specifies exactly what
Claude should return. Request ten labeled sections,
the viewer, the promise, the angle, five title options,
five thumbnail directions, three opening hooks, a
section by section outline, examples or evidence to include, obviously, and then claims
that require verification, and finally, questions
that remain unresolved. Now, notice that the
prompt doesn't ask Claude to make one final
decision for me, right? It asks for options and
an organized draft. I still want to evaluate the results as the
creator myself. This is just kind of
piggybacking off of the AI and making sure that
AI is doing the work for me. Now, I'll submit the
prompt, and I'll let Claude build a creative brief. And when the document appears, I'll confirm that the requested sections are obviously present. At this stage, I'm not
polishing every sentence, okay? I'm just checking whether Claude followed the structure and whether or not the response is grounded in the decisions
that we've already made. Now, the four part prompt is obviously reasonable
beyond this example, which is extremely valuable. Context is going to tell
Claude what it can trust. Task defines the job, constraints protect
accuracy and tone. Output defines the deliverable, and if the result is weak, you can usually improve it
by strengthening one of these four parts instead of starting over with a
completely different prompt. Now, by the end of this step, we've moved from
scattered notes to a structured creative
brief that we can now evaluate
package and refine.
4. Plan Packaging and Outline: Brief now has a
strong foundation, so we can move directly into the packaging and delivery now. We're going to use
Claude to develop accurate title options, visual thumbnail directions,
and an opening hook, as well as a section
by section outline. Now, more importantly,
I'll show you how to evaluate those ideas instead of automatically accepting the first response
just from the get go. Now that Claude's
created the brief, we're going to evaluate the
packaging options and turn the best ideas into a
practical video outline. Packaging means the title, the thumbnail direction,
and the opening hook. These elements should
make the lesson clear and interesting without
promising an outcome that the video can't guarantee. Now, I'll begin
with title options. Claude has returned
five approaches rather than five nearly
identical rewrites, and that's a different
distinction. One title explains
the process directly. Another challenges
a common habit. A third focuses on the problem with automatically
selected highlights, and another states the exact
skill in plain language. Now, the final option uses a rejected clip and useful
clip comparison as well. Not choosing a title simply because it sounds
exciting, right? I compare every option
with a confirmed promise. Does the title accurately describe what the
student's going to learn? Does it suggest views, virality, or growth that
the lesson does not teach? Does it sound like something it can genuinely demonstrate, that I can actually
genuinely demonstrate? And for this example,
a direct process title is the safest starting point, in my opinion, because it names the transformation without
exaggerating the end result. Now next, I'll review the
thumbnail directions. Claude's got a suggested
side by side verdict here and a transcript
turning it into three clips. Now, a giveaway phrase
with a strike through, a four part checklist, and an auto selected clip
that still needs review. These are directions.
They're not finished graphics, of course. I'm evaluating the
visual behind each one. A strong thumbnail should communicate one concept quickly. It shouldn't try to display the entire workflow,
in my opinion. The side by side direction
is useful because the lesson includes a rejected example and a useful example, as well. Now, the checklist
direction is also strong because it represents the
four criteria visually. Whichever direction I do choose, I'm going to use my own
footage rather than displaying a recognizable third
party interface as the main visual there. Then I'm going to move
into the opening hooks. So the first hook starts with a cold demonstration
of a rejected clip. The second names the
habit of choosing the most energetic moment or accepting an
automatic selection. The third leads with
a core principle that good shorts is a
complicated idea. Now, the best hook depends on the footage that I
actually process, of course. A cold demonstration is
powerful only if I have a real rejected clip that
clearly shows that problem. I don't, then the principal led opening is probably more
honest and easier to support, and this is why we treat Claude suggestions
as options, right? Just that, suggestions
rather than actual facts. Now, I'm going to review the
section by section outline. So the structure
begins with a hook and introduces the
four part test here. It then shows how to review the transcript before cutting. After that, it's
going to compare a rejected clip with a useful one, and the next section
restructures the selected moment
around one promise. Then it's going to
prepare the clip for vertical viewing and finish
with a standalone review. Now, I'm going to check the sequence for
cause and effect. The student first
needs the test, and then the student
needs to see why a clip fails, of course. Only after that does restructuring
actually make sense? The order should match the way that somebody
would perform the skill, not merely the order in which the ideas appear
during brainstorming. I'll also look for places where one section is doing
too much, right? The restructuring section should stay focused on one promise, a stronger opening line, and remove unnecessary setup. Detailed editing instruction is outside of the scope
of this lesson. So the outline mentions vertical reframing and readable
captions without being, without being a full
editing class entirely. Now, finally, I'm
going to compare the packaging with the outline. So the title, the
thumbnail, the hook, and the lesson structure should all make the same promise. If the thumbnail
emphasizes one AI tool, but the outline is
like tool neutral, then obviously that's going
to create a mismatch. Now, if the title promises
three finished clips, but the demonstration
only shows two examples, that needs to be resolved
before recording as well. At the end of this
step, we're going to have more than
a list of ideas. We're going to have
a packaging options that have been tested against the source material of
an outline and obviously arranged around the actual skill that the student will perform.
5. Refine and Verify the Brief: AI generated work can sound polished while
still being generic, repetitive and unsupported
in a lot of cases, right? So in this lesson,
we're going to use a final review process
to remove weak material, separate facts from
assumptions, which is huge. We want to verify the
information that we have. We want to flag anything
that needs a source, and we want to
convert the work into a clean one page brief
that's easy to digest. In this final lesson,
we're going to audit the creative brief
before treating it as production ready. Now, the step
separates supported information from creative
recommendations. It identifies claims
that need evidence and turns unresolved assumptions
into clear questions. Now, I'm going to
return to Claude here, and this conversation
is going to ask for an audit of the brief. Now, my prompt
requests for lists, statements directly supported
by my source notes, reasonable creative
recommendations, claims or factual statements that
require an external source, and generic or repetitive
material that can be removed. Now, I'm also going
to tell Claude not to verify anything from
memory and don't invent sources just
to make sure that it's obviously,
staying accurate. The first list shows
what's directly supported. This is going to include the viewer description,
the promise, the four criteria for
a completed clip, and a tool neutral scope. And the rejected verses useful comparison is going
to be helpful as well. Now, these elements came
from the confirmed notes, so they can remain central
to the lesson, of course. The second list identifies
creative recommendations. So title, thumbnail,
directions, opening hooks, and the exact ordering of the outline are obviously
recommendations. No they might be good ideas, but they're not facts
from the source material. So labeling them this way
keeps us from confusing the AI suggestions with our
own established process. Now, third, the list contains claims that
need checking, right? So, for example, saying that every step can be completed with free tools is factual claims about the current
product features. That needs to be
verified against the specific tools shown at
the time of the recording. Saying that AI selected
highlights can lack context, and saying that AI selected
highlights can lack context should be presented
as an observation supported by a real example, not universal facts about
every single tool, of course. Now, I'm also going to
check first person wording. A title such as, you know, I Choose short
clips implies that I have actually used
this process already. And a hook that calls
the rejected clip a real moment for my own
tutorial might also be true. Now, Claude can write convincing
first person language, but only I can confirm whether it actually
represents my experience. Next, I'll look for
repetitive material. So the brief repeats
the warning about performance claims
in several places. One clear statement is enough. Filler, dns, and repeated versions of the four part test can
also be removed, too. Cutting repetition
makes the Brief easier to use while
recording, too. Now, the unresolved question section
is especially valuable. It asks what the rejected and useful clips
actually contain. Whether the video
is going to show three finished clips
or only two examples, whether real AI selected
highlight is available, whether a new opening line
will be demonstrated, and which section specifically belong in the one
page class project. Now, these questions
aren't failures, right? They're a production checklist. Each unanswered question
points to a decision, an asset or a demonstration that's got to be
prepared during filming. Now, I'm going to
finish by making three passes through the brief. In the accuracy
pass, I'm going to remove or qualify
unsupported claims. In the usefulness pass, I'm going to keep
recommendations to make the lesson
easier to understand. In the production pass,
I'm going to confirm that every promise example in visual can actually
be shown on screen. And the final brief should be specific enough to
guide the video, but honest about
what's confirmed, recommended or still unresolved. And that's the difference
between using Claude as a writing shortcut and using it as a structured
creative partner.
6. Class Project and Next Steps: Alright, so you now have
the complete process, so now it's time to create
your class project. You want to choose one
YouTube video idea and gather a small amount
of source material, right? That can include your
personal experience, notes, examples, research, or a transcript from something that you've
already created. Then you want to
use the five part creator brief to
define the viewer, promise, angle,
packaging, and delivery. And your final brief
should include three title options, three
thumbnail directions, and an opening hook, a section by section outline, and a verification section that identifies any claims
that still need a source. For the project gallery, upload a completed brief as a PDF, an image or a
document screenshot, so everybody can see it, and I can obviously
give you feedback. And if any information
is obviously private, then replace it with
neutral examples or blur it out or obviously color over it with paint
or something like. In your project description, tell us which title and
thumbnail direction that you selected and explain why they accurately
represent the video. And as an optional stretch goal, use the brief to write
the first 60 seconds of your video with
your own voice. The Creative Brief template, the Pompact and the
Review checklist are all included in
the class resources. I want to thank you
for taking this class. I'm looking forward to seeing the video ideas
that you develop. And if you put all this together and take action on
it consistently, that's when the real
results start to show up.