Heads up! To view this whole video, sign in with your Courses account or enroll in your free 7-day trial. Sign In Enroll
Well done!
You have completed No-Code Prompt Engineering!
You have completed No-Code Prompt Engineering!
Preview
Explore prompt engineering for no code development by enhancing your AI coding skills in this interactive workshop. Learn to craft clear, powerful prompts that enable AI tools to work smarter and faster, transforming your no code development process. Whether you're new to AI or already coding, you'll discover actionable strategies to streamline your workflow, solve challenges more efficiently, and achieve superior results from every prompt. Join us for a hands-on experience that will elevate your AI coding game and empower your no code projects.
This is part of the AI for Everyone: Build Power Apps Without Coding track. Check out the track to see the rest of our No-Code content. Stay tuned—new content is continually being added, so be sure to check back regularly for updates!
Tools used in this video:
Documentation:
Related Discussions
Have questions about this video? Start a discussion with the community and Treehouse staff.
Sign upRelated Discussions
Have questions about this video? Start a discussion with the community and Treehouse staff.
Sign up
Hello everyone, and welcome back
to our series.
0:09
It's Dan and Christian here yet again.
0:12
And in this video we would like to discuss
how to get the most out of your prompts,
0:14
aka maximizing your prompts,
getting the most out of it,
0:18
and making sure that you're not confusing
or overwhelming the AI,
0:21
but you're also giving it
enough information
0:26
that it understands you completely,
0:28
and it's not running off on a tangent
and building its own creation.
0:30
And our first point here is be clear
and precise.
0:33
You want to give the AI clear directions
with your prompt,
0:37
and you don't want to be too vague or have
the AI fill in too many of the blanks.
0:41
So, for example, here
I went to perplexity AI,
0:46
and in this example I just clearly said,
can you show a list of three examples
0:50
before and after of being clear
and precise when crafting
0:55
a prompt for coding AI,
and it gave us a few examples.
0:58
Example
one is sorting the list of integers,
1:03
so I may, you know, jump into lovable
1:06
or replit and may ask the
AI to write a function to sort numbers.
1:10
It may get what I'm asking for,
but it may not be enough information.
1:15
So Perplexity suggests a clear and precise
version of this prompt would be.
1:19
Generate a Python function
that sorts a list of integers
1:23
in ascending order
using a quicksort algorithm.
1:27
Now, the first thing you may be
thinking is
1:30
there is a lot of technological
or a lot of, tech terminology right here.
1:33
If you're a non-technical creator, user
or founder, this may be a little off
1:38
putting to you, but with AI,
we have the ability to not only get inputs
1:43
and get outputs, but to ask questions
and to go back and forth.
1:49
It's very conversational,
so I would suggest asking Perplexity
1:53
or ChatGPT or even Claude to explain
1:58
these tech terminologies
or these tech terms in greater detail
2:02
if you don't know them, to provide
a list of them, if you don't have them.
2:05
So it can give you anything you need
for context.
2:08
What you can even do is,
my favorite kind of direction to give,
2:12
any of these kind of chat
2:17
programs is explain to me like I'm five,
what you're talking about
2:20
in this statement, and it will list out
exactly what you need to understand.
2:25
And then from there, just talk with it
and kind of refine
2:30
what you're trying to think about.
2:33
And at the very end, it's a great
thank you for this.
2:35
Can you now
rewrite a clear, concise statement
2:38
that I can give X program, Y program
to do what you need to do?
2:41
And one last thing is
you want to think of it as
2:46
if you're talking to a stranger
2:49
and you tell the stranger, hey,
can you sort this list for me?
2:51
They're not going to know what you mean.
2:54
So think of chatting with ChatGPT
2:56
or Claude when you start a new one,
as if you're talking to a stranger.
2:58
You need to give just enough context
or a little bit more for it to understand.
3:02
Okay, you know, Christian is talking
about building a sorted list using Python.
3:06
Great. Now you know what to do.
3:12
Absolutely. That's perfectly correct.
3:14
It's a build on that.
3:16
Just always think of your agent, your
AI companion, as a intern.
3:17
You know, first day on the job.
3:22
This person you know is willing to
help is right.
3:24
They're they're here for you,
3:26
but they might not necessarily know
exactly what you want.
3:28
So overexplaining can always, you know,
3:31
go above and beyond to help
get the results that you desire.
3:34
Now next up, we want to make the point
to use AI to improve your prompts.
3:37
And next time we're going to start
a new chat here in Perplexity.
3:41
And I'm going to just show
some examples of
3:44
how you can use AI to improve your prompt.
3:47
Now, before I simply just go in
and say something like,
3:50
I want to create a to do list app.
3:55
Help me create a prompt
4:02
that AI would understand
4:05
to generate
4:09
this.
4:11
Now I could clearly do this
and just send this in.
4:12
And matter of fact, we will go ahead and
do that just to see what the results are.
4:15
But as we just said, it's kind of best
if you can use some of that terminology.
4:19
So we'll go back
and see how we can restructure this prompt
4:24
after we see what Perplexity,
you know, kind of creates for us.
4:27
So we have to create
a prompt needs to be specific
4:32
about the features and functionalities,
which I totally agree with.
4:37
So let's see the exact prompt.
4:40
It's design a comprehensive to do list
app with the following features.
4:41
Task management allow users to create,
edit and delete says
4:45
include options for setting due dates,
reminders, and task prioritization.
4:49
Organization.
4:53
Provide features for categorizing tasks
into projects or contexts.
4:54
That's awesome. Collaboration.
4:57
Enable users to share tasks with others
and assigns has to team members.
4:59
This is a good feature,
5:03
but if we were building this out in real
time, I would probably just delete that.
5:04
I don't think we actually need
collaboration.
5:08
That's a little bit more of an advanced
feature customization offer customizable
5:10
views, filters, and notifications
to fit individual workflows.
5:15
And that's okay as well.
5:19
And, you know, just
you can continue to read on some of these.
5:21
Hit the mark there. Perfect.
5:24
Some of them, they're a little bit
more advanced for what we may want.
5:25
So we would kind of take that out
5:29
and we can just take it out
and copy and paste and use as we like.
5:31
But honestly, this section right here
5:34
is a very strong starting point
when wanting to build.
5:39
The only thing I would suggest is
if you had
5:43
specific technology
that you wanted to build with in mind.
5:47
So you had coding languages like Python,
or you had specific
5:51
libraries or dependencies that you knew
you wanted to work with.
5:55
So you knew you wanted,
a Supabase database, for example.
5:59
But you can include that in there.
6:03
And, you know,
6:04
we could even ask for a follow up
prompt like, please improve this prompt.
6:05
By including
6:12
the tech stack.
6:14
The programing language
6:19
and any libraries or dependencies.
6:22
And again, I didn't tell it specifically
6:29
that we wanted
to use Python or JavaScript.
6:30
So we're going to see what Perplexity
recommends in this example.
6:33
So for the front end
6:38
Perplexity suggesting that we build
the user interface using react
6:41
with TypeScript utilizing the material
UI for styling components,
6:44
the backend is suggesting Node.js
with express JS.
6:49
The database is a MongoDB
mongoose database,
6:53
and that's using the ODM object data
modeling library.
6:56
And again, you don't have to be
very familiar with these.
7:00
Like for me, I'm not extremely familiar
with the mongoose database.
7:03
So actually let's take this.
7:08
And just as Christian suggested.
7:10
Explain the above to me.
7:15
Like I was five years old.
7:20
Now, while Dan's doing that
7:24
when you're looking at
what it's typing out for your tech stack.
7:26
Tech stack is basically all the necessary
kind of programing languages
7:30
or databases that is required
for your application to run.
7:35
You might get overwhelmed with,
oh my gosh, what's Node.js?
7:39
What's swift? What's this? What's that?
7:44
That's 100%
okay to feel overwhelmed at first.
7:47
Take your time.
7:50
Talk to it like what Dan is saying.
7:51
Like, hey, explain to me like I’m five,
what does Node.js do?
7:53
So okay,
so you can kind of get a gist of what each
7:57
component of your app
that's going to be built does.
8:00
That's
going to save you a lot of headache.
8:04
Once you go into more complex additions
8:06
to your app that you're creating 100%.
8:10
So like in this example,
we asked what is MongoDB?
8:14
And it's breaking it down that it's kind
of like having a big box full of folders.
8:17
And each folder can hold lots of papers,
and each paper can have different kinds
8:21
of information on it.
8:25
So that's just breaking down
how the database works.
8:26
And then it even goes on further
to understand or
8:29
explain that MongoDB is the database
itself.
8:32
That's where you store all your documents.
8:35
And then mongoose is a helper tool,
that makes it easier to work with MongoDB.
8:37
So just this little bit of explanation
8:42
I kind of understand what MongoDB is,
what mongoose is.
8:45
And how that relates to MongoDB.
8:49
And then kind of the structure
of how this database is going to work now.
8:51
But let's, you know, take it
a little further and let's just say
8:55
explain node JS
8:58
and why I should use it for this project.
9:02
And while
9:06
Dan is waiting for that
to kind of populate,
9:07
if the examples that pop up
doesn't make sense to you,
9:11
provide
an example that you would understand.
9:15
So like for the MongoDB mongoose,
I would say, hey,
9:19
I don't understand this example.
9:23
Can you provide it in a context of,
let's say a teacher and a helper,
9:24
or can you provide in a context of a baker
9:28
things that you kind of have
a understanding of
9:31
a will providing example like that, so
you can understand it from that context.
9:35
So that you can bridge that
gap of technical to non-technical jargon.
9:38
Absolutely. That's a great tip Christian.
9:43
Here we have our completed, answer.
9:46
And so what is Node.js is simply says
Node.js is an open source,
9:48
cross-platform
JavaScript runtime environment that allows
9:52
developers to run
JavaScript on the server side.
9:56
Okay, so it's a little technical
and we could ask for it to break that down
9:59
even further.
10:02
But basically it's just saying
10:03
it's an open source platform
that allows us to run JavaScript.
10:04
And then some of the key features are it's
asynchronous and event driven.
10:08
So Node.js doesn't wait for operations
like a database query or files.
10:13
It's not waiting one by one.
10:17
It's doing these tasks at the same time.
10:19
So that's pretty cool.
10:22
And then there's cross-platform
compatibility.
10:23
So Node.js can run on windows,
Mac OS and Linux.
10:25
So it's versatile.
10:29
So that right
there lets us know that that's
10:30
probably a good technology
to have for any of our projects,
10:32
because it gives us instant
platform compatibility.
10:36
We’re not having to refactor or redeploy
or rebuild
10:40
this app for these different platforms.
10:44
Scalability.
10:48
Node.js is highly scalable.
10:49
It allows it to handle growing traffic
without significant performance drops.
10:51
Again, that's going to be great
for our projects too, where
10:55
whether we start with just a few users
or it grows over time and we,
10:58
you know, scale it and becomes
a paid product or something bit larger,
11:02
we, you know, can handle that scale.
11:06
And then it's perfect for full stack
JavaScript development.
11:09
Node.js enables developers to use
JavaScript on both sides, both server
11:13
side and client side, reducing the need
to learn multiple programing languages.
11:17
So awesome.
11:22
That just helped us understand
a little bit more about Node.js.
11:22
And again, we don't have to know it
front to back or understand
11:26
it as an expert,
we can continue to ask questions
11:30
to, prompt Perplexity, ChatGPT, Claude,
11:34
whomever to give us more context
and understanding and to make it
11:38
more relatable to our specific use case
or our background.
11:42
Up next, I would like to talk about
improving your prompt with AI
11:46
and using it to go beyond
just adding the technical terms
11:49
that you may not know, but having it
help you generate a prompt entirely.
11:53
If you just have a round about ideal
11:57
or how to kind of improve the prompt
you already have.
11:59
So let's head over to chat GPT.
12:03
And for example, let's just say
we wanted to make a basic to do list that.
12:05
So I'm going to put I want to create a
12:09
basic to do list that
12:12
will help me
12:17
improve this prompt.
12:21
And optimize it
12:25
so that the AI understands my request
12:28
and can help
12:33
build it out fully.
12:36
And right now we're using one of the newer
12:45
ChatGPT models 03-mini-high.
12:47
So we'll just give it a second
to kind of think about the task at hand.
12:51
And then we get our prompt.
12:56
So again our original prompt
12:58
was I want to create a basic to do list
at which I didn't even spell correctly.
13:00
As you can see here.
13:04
But when you go down here,
13:06
here's what we have is a new prompt.
13:09
I want to build a basic
13:11
to do list application with both front end
and back end components.
13:12
The app should allow users to perform
the following actions.
13:17
Create.
13:20
Add new tasks, read, view a list of tasks,
update edit
13:21
existing tasks and mark them as completed
and delete remove tasks.
13:26
And this is also known as CRUD.
13:31
Being able to create,
read, update and delete.
13:33
And also it has here that we'd like to
implement the app as a full stack project.
13:37
So for example, use a framework like react
or another of your choice on the back
13:41
end is suggesting create a RESTful API
using a framework such as express
13:45
JS, flask,
which is a Python framework, or Django
13:49
which is also Python to handle
and store the Crud operations.
13:54
So that just means anytime
a user is creating a new task,
13:57
reading a task, updating an existing task,
deleting all these actions
14:02
while being performed by the user on
the front end are being recorded
14:07
and updated on the back end,
and it's using one of these tech stacks
14:10
or one of these technologies
that we choose to perform that.
14:14
And then just data
storage was just having a basic database
14:18
so that we can have an appropriate method
to store everything.
14:21
And then it's just provide, step by step
guide on setting up the project.
14:26
Our recommended folder structure
code examples and explanations
14:31
of best practices for error handling,
scalability and maintainability.
14:36
Now just personal preference.
14:41
I feel a couple of these last requests
are just a little extra.
14:43
Just like a recommended folder structure
when building with the
14:48
AI is going to kind of build
all that yourself.
14:51
Replit shows you its folder structure
and then, something like lovable.
14:54
You won't see the code structure
until you deploy the code
14:59
to like, GitHub, but it it kind of handles
that on that end.
15:02
And then code examples
we don't really need to see as well.
15:05
But I would say if you wanted to,
you could definitely prompt
15:08
those eyes
with those to see what it gives you.
15:12
But even more
so you could ask for that here in ChatGPT
15:14
because again, ChatGPT
has the ability to produce code as well.
15:19
So it could give you some examples
that you might like,
15:22
and then you could take that
as further examples with your prompt
15:25
to bring all that into Lovable,
Replit, Bolt.new,
15:29
or whatever tool you're using to provide
that extra context.
15:33
Because personally,
I like to provide as much
15:37
context as possible,
and that includes the actual
15:39
prompt itself on what we're building
and what we're doing.
15:42
But also an example.
15:45
So it knows kind of
what we're looking for,
15:47
whether that example is a code sample
or an image.
15:49
If you don't know what correct code
looks like or what ChatGPT is generating
15:52
as it has good code, you can also just
providing your prompt, I will.
15:58
I can make a to do list app list,
whatever the prompt is.
16:03
Use Todoist as a exemplar model
16:07
and it will know what that means.
16:13
Obviously provide a link for the Todosit
and then it will scrape that website
16:15
to see what it is about.
16:19
And then furthermore,
16:21
if you feel intimidated or that it's
getting a little too much detail,
16:23
you could always ask
to simplify this prompt,
16:27
whether that helps you understand it
better and get more comfortable with it,
16:30
or whether you feel that
16:33
that this prompt is too advance
and it might trip up the AI.
16:34
So always feel free to do that
16:37
and lastly, in regards
to using AI to improve your prompt.
16:39
I'd like to touch on a tool by Anthropic.
16:43
Anthropic has the anthropic console.
16:46
Now, this tool is mainly made
for developers and builders.
16:48
It's so you can generate API keys and
get those and add those to your projects.
16:52
And you would want to get an API key
to add to your project.
16:57
So you can add chat completion
functionality.
16:59
So if you built an app that you wanted
Claude to provide feedback and answers
17:03
through the app,
you would have to get an API key first.
17:08
But that's not here nor there.
17:12
What I really want to talk about
are these simple features right here.
17:14
Generate a prompt
and improve an existing prompt.
17:17
Now, once you sign up for the console,
you have an account you're logged in.
17:20
You can simply click generate a prompt.
17:24
So again
I'm going to go right back to chat GPT.
17:27
And I'm going to copy our original prompt.
17:32
Before we asked for the extra help
17:35
I'm going to paste.
17:39
I'm going to fix that spelling error.
17:40
And I'm going to hit generate.
17:42
And Claude
is going to just take that simple
17:45
task and break down a full
17:48
prompt that we can work with here.
17:52
So it says you are tasked
with creating a basic to do list app.
17:55
Your goal is to design
a simple, functional application
17:58
that allows users to manage your tasks
effectively.
18:01
Follow these instructions
to complete the task.
18:05
Review the list of features.
18:07
So here's where we can put
in all the features we want.
18:09
Consider the target.
18:12
Here's where we could
consider the target platform,
18:13
and then design the basic structure
of the functionality to do list app.
18:17
Keep it in mind. The following guidelines
18:20
focus on implementing the core features
listed above.
18:22
Ensure
the app is user friendly and intuitive.
18:26
Consider how the app will function
on the specified platform.
18:29
Keep the design simple and minimalistic
18:33
and furthermore, like I said,
18:37
anything can be tweaked, changed, deleted.
18:39
If for whatever reason I think
keep the design simple and minimalistic
18:42
is a great tip to have.
18:46
I can delete that from here,
but you know I'm going to undo that.
18:47
But here we could say features.
18:52
So just going off that last
ChatGPT prompt,
18:54
we want the ability to create new tasks.
18:58
We want to be able to update
19:04
and view created tasks.
19:07
We want to be able to delete tasks.
19:13
And are there any other features
19:19
you think we might need, Christian?
19:21
I think
19:25
we have everything right.
19:27
CRUD Create, read, update and delete.
19:28
Yep yep. Update.
19:33
Yep. Create.
19:34
And I'm just do create
and we'll have update and view.
19:36
So that's you know update and read.
19:39
So yep we have everything.
19:41
And then consider the
19:43
target platform for the app.
19:44
I'm not sure exactly what it means
by this.
19:47
I think it means like
19:51
if it's going to be on iOS, Android, web,
that's great.
19:53
Thank you for helping me with that.
19:58
In that case, let's just say
19:59
this is going to be a web app.
20:02
That works across
20:10
all systems and browsers.
20:12
So again, right there,
20:19
we have a pretty good, prompt.
20:21
It's a lot more detailed and a lot more
feature rich than what we originally had.
20:24
Again, I could simply give Lovable, Bolt,
whatever we use.
20:30
I can simply give that a create
this To-Do list app
20:35
and it will get to work,
but your results will be hit or miss.
20:39
It'll be completely different. Ever.
20:44
Every time we could get like a simple,
minimal basic To-Do list app
20:45
with just the features
that we were thinking of,
20:50
or we get something really complex,
or we get something
20:53
that you can only add
one task at a time and cross it off.
20:56
It might have like a timer for us to add
the test to the timer and, you know,
21:00
set a Pomodoro timer for how long
we want to have to do that task.
21:05
But again, that might be out of scope.
21:09
We might not have wanted that.
21:11
That might have just been a feature.
21:13
It just kind of came up with on its own.
21:14
So that's why getting more structured,
getting more detailed
21:16
and becoming a little bit
more nuanced in what you want
21:20
and what you don't want
is a much better way to go.
21:23
And lastly, what I would like to suggest
is providing an outlet of your ideal
21:26
to the ChatGPT or, you know, to your,
tool that you plan on using,
21:31
but also taking this step further
21:36
and providing a PRD
de a project requirements doc.
21:39
This kind of lets the I know everything
that the project encompasses
21:43
and is made of,
and also allows it to think
21:47
long term, so it'll know to build features
21:51
with other features at by and saving space
21:54
and saving room to integrate
and incorporate everything together.
21:57
Now one warning or one guardrail
I would like to bring up about when using
22:02
PRDs is to provide the PRD
22:07
like paste it
22:12
in, but then after you paste it in,
22:13
include another paragraph or sentence
22:16
that says okay, that's the entire project.
22:20
Now that you understand
what we were looking to build,
22:23
please start with and then describe
one aspect of the project.
22:26
So like for this to do list, for example,
let's give it
22:31
the project requirements document
if we have one made.
22:35
But at the end say now focus on
generating the UI of a blank screen.
22:39
Allow us to add our first task.
22:45
Then we can
of course incorporate new features
22:47
like updating tasks, deleting tasks,
adding user settings, user
22:51
authentication On top of that initial UI,
we built out.
22:56
The PRD.
23:00
The project requirement doc is a great
tool for the AI to always refer back to,
23:01
because the longer your chat is,
the more you're chatting with it.
23:05
It can get lost in his chain of thoughts.
23:09
That's why when you have a PRD document
saved on your project, file structure,
23:12
you can always refer to it
23:19
like we just finished feature one,
which is building out the UI.
23:20
This focus on feature two
now refer to the PRD document for what
23:24
feature two is and continue from there.
23:28
So it keeps the AI on track
instead of assuming
23:30
from your previous list of conversations
what it is supposed to do.
23:34
So this not only prevents hallucination
23:39
or the
AI kind of just spitting up what it wants,
23:42
it allows the AI to have the quote
unquote, freedom to build something
23:47
within the constraints of your project
requirement document the PRD.
23:53
Absolutely.
23:59
And and if you don't understand,
you know, exactly
24:00
what a PRD or our project requirements
document is.
24:03
No worries.
24:06
We have another video covering that,
and it'll be linked down in the teacher's
24:07
notes below.
24:10
So feel free to watch that video
and get more information on the PRD.
24:11
Now. Moving on.
24:16
The last thing I want to touch on
when using Loveable boat
24:17
or anything for building
and you want to improve
24:22
your prop is simply just check
with the platform's documentation.
24:24
Lovable, Bolt, Replit.
24:29
They all have documentation sites
that are fully equipped
24:31
with best practices,
prompt examples, and more.
24:35
Matter of fact, for example,
let me hop into prompting with lovable.
24:38
So this here is a user guide.
24:42
It's a part of the documentation
here at Lovable.
24:44
And you can find it at docs.lovable.dev.
24:47
And let me just make
sure we're zoomed in here.
24:51
There we go.
24:55
Make that a little easier to read.
24:56
Now this document,
while is specifically tailored to lovable,
24:58
you'll be able to pick up on some best
practices that you can use with any tool.
25:04
And that's the common thread
I have, the Replit doc here open as well.
25:08
And whether you're using Replit,
whether you're using lovable,
25:13
you can find a few tips
that could apply across the board.
25:16
But back here at lovable, for example,
it just gets into the basics
25:21
of where to prompt,
but then it also gets into prompting
25:25
strategies like contextual prompting.
25:28
So this is, in the discussion of wanting
to build a project management tool.
25:31
It gives this, detailed, simple prompt.
25:37
It says we are building
25:40
a project management tool that helps teams
track their tasks.
25:41
This tool should have features
like user authentication
25:45
on project creation, task
assignments, and reporting.
25:48
Now for the first task, create
the UI for a project creation.
25:52
So again, that kind of goes back to what
we just talked about.
25:56
It gave an overview of what
we're looking to build, but
25:59
it broke down specifically to the AI.
26:02
Please
start with this one part and work on that.
26:05
So then we can stack all of that further.
26:09
And again, another example
that provides here is I need a CRM app
26:11
with Supabase integration
and a secure auth flow.
26:15
Start by setting up the back end.
26:18
So again
this is even a little less detailed,
26:21
but it still says what
we're looking to build and where to start.
26:23
So again, that's just great context.
26:28
It has other
26:31
examples like here what not to do.
26:31
So here the user put in
build a CRM app with Supabase
26:35
Auth, Google Sheets, export,
data enrichment.
26:38
It's just all listed together
and you know, very high overview,
26:42
just the name of what they want it.
26:45
Whereas here
it's a little a little bit more detailed.
26:48
It says, you know, do
set up a Supabase connected
26:51
CRM back end and a secure
all flow with user roles.
26:54
Integrate Google Sheets to export records.
26:59
So again,
instead of just Google Sheets export,
27:02
get a little bit more detailed,
explain the functionality and why.
27:04
And that just gives more context to the AI
27:08
and helps it understand in greater detail.
27:12
And again, we're not
27:15
going to go through this entire page,
but you can go through here.
27:16
Here's best practices
for using images in your prompts.
27:20
Feedback
27:24
you know
make sure you're avoiding ambiguity.
27:26
So again make this that better.
27:28
That can be so many things. Yes.
27:31
Thank you.
27:34
That could be so many things
that make that better.
27:35
Could be new features that could be making
the functionality better,
27:38
that could be improving on the layout,
what the user actually wanted here.
27:42
Refactor the app to clean up
unused components, improve performance.
27:46
So and without changing the UI.
27:51
So basically what they wanted
was to improve
27:53
the speed of the app
by cleaning up unused code.
27:55
Totally different
than just make this better. Yep.
27:59
And a lot of times
when you write, make this app better,
28:03
it will make something better
by breaking something else.
28:07
So that's one of the main frustrations
with these tools
28:10
is that if you are vague
with what you want, it's going to,
28:14
with high probability, break
something else that was working before.
28:19
So the more details you can provide,
like the example provided here,
28:22
the easier it is for you to understand
when and where things are being changed.
28:29
And remember,
this is one tool in your whole workflow.
28:34
So if you don't know what to write, go on.
28:38
ChatGPT is like, hey,
28:42
here's the documentation from Lovable
that says Best Practices for prompting.
28:43
I'm trying to do this
using their guidelines.
28:48
Help me write a better
appropriate prompt for it.
28:51
Then it will generate that
like you will start developing
28:55
these habits of double checking
28:59
with different tools
to ensure that whatever prompting
29:02
you're doing aligns
with the output that you want.
29:06
Absolutely.
29:11
That was perfect.
29:12
Adding that context like the documentation
straight linking that to GPT
29:13
when making the prompt,
that's kind of one of the best ways to go.
29:18
So it has that clear context.
29:22
So it's building
the prompt for the tool you're using.
29:23
Perfect.
29:27
And I just wanted to add that the links
to the lovable documentation,
29:28
this Replit documentation,
all of this will be included
29:33
in the teachers notes below,
so you won't have to go searching for it,
29:36
so feel free to check that out.
29:39
But, I think that's going to do it
for this video.
29:41
I think that helps give you a greater
understanding of how you can get
29:44
the most out of your prompts.
29:48
And we'll be back with another video
on, No-Code and using AI.
29:49
We'll see you in the next one.
29:54
Thanks, y'all.
29:56
You need to sign up for Treehouse in order to download course files.
Sign upYou need to sign up for Treehouse in order to set up Workspace
Sign up