Prompt Engineering Lifecycle and Tools
Key Points
- Prompt engineering is a “wild‑west” space that’s become essential to AI workflows, yet few have mapped out a systematic prompt life‑cycle.
- The first stage—authoring and drafting—relies on interactive tools (Claude, ChatGPT, Prompt Perfect, Cursor) to iteratively refine wording and clarify mental models.
- After drafting, prompts are version‑controlled like code, using platforms such as PromptLayer, PromptMethus, git‑based repos, or LangChain to ensure persistence, auditability, and team coordination.
- Production‑grade prompts then enter a testing phase where multiple versions are evaluated for accuracy, cost, and hallucination risk, using tools like PromptFlow, Hegel’s evaluation suite, or custom eval frameworks.
Sections
- Systematic Prompting Lifecycle Overview - The speaker outlines a structured life‑cycle for crafting LLM prompts—detailing stages like authoring, drafting, and versioning—and highlights the specific tools suited to each phase.
- From Prompts to Agentic Workflows - The speaker explains how prompts evolve into guided, multi‑step agents that rely on frameworks, tooling, and production‑level safeguards such as traceability, safety, and governance.
- Prompt Ideation with Hey Presto - The speaker highlights the common failure to define output formats during early prompt drafting, introduces the Hey Presto tool designed to guide fuzzy intent formation and format‑specific prompt tuning, and demonstrates its use for creating code prompts for a travel app.
- Auto-Generating Slide Decks with Hey Presto - The speaker shows how a single prompt in Hey Presto instantly creates a polished presentation, highlights seamless integration with Claude and ChatGPT, and candidly addresses audience skepticism about product promotion.
- From Idea to Intent: Hey Presto - The speaker outlines the need for a simple, community‑driven tool to transform casual ideas into precise AI prompts, describes building Hey Presto to fill that gap, and invites feedback and collaboration.
Full Transcript
# Prompt Engineering Lifecycle and Tools **Source:** [https://www.youtube.com/watch?v=V0YhpeSOuzk](https://www.youtube.com/watch?v=V0YhpeSOuzk) **Duration:** 00:16:41 ## Summary - Prompt engineering is a “wild‑west” space that’s become essential to AI workflows, yet few have mapped out a systematic prompt life‑cycle. - The first stage—authoring and drafting—relies on interactive tools (Claude, ChatGPT, Prompt Perfect, Cursor) to iteratively refine wording and clarify mental models. - After drafting, prompts are version‑controlled like code, using platforms such as PromptLayer, PromptMethus, git‑based repos, or LangChain to ensure persistence, auditability, and team coordination. - Production‑grade prompts then enter a testing phase where multiple versions are evaluated for accuracy, cost, and hallucination risk, using tools like PromptFlow, Hegel’s evaluation suite, or custom eval frameworks. ## Sections - [00:00:00](https://www.youtube.com/watch?v=V0YhpeSOuzk&t=0s) **Systematic Prompting Lifecycle Overview** - The speaker outlines a structured life‑cycle for crafting LLM prompts—detailing stages like authoring, drafting, and versioning—and highlights the specific tools suited to each phase. - [00:04:11](https://www.youtube.com/watch?v=V0YhpeSOuzk&t=251s) **From Prompts to Agentic Workflows** - The speaker explains how prompts evolve into guided, multi‑step agents that rely on frameworks, tooling, and production‑level safeguards such as traceability, safety, and governance. - [00:07:19](https://www.youtube.com/watch?v=V0YhpeSOuzk&t=439s) **Prompt Ideation with Hey Presto** - The speaker highlights the common failure to define output formats during early prompt drafting, introduces the Hey Presto tool designed to guide fuzzy intent formation and format‑specific prompt tuning, and demonstrates its use for creating code prompts for a travel app. - [00:11:13](https://www.youtube.com/watch?v=V0YhpeSOuzk&t=673s) **Auto-Generating Slide Decks with Hey Presto** - The speaker shows how a single prompt in Hey Presto instantly creates a polished presentation, highlights seamless integration with Claude and ChatGPT, and candidly addresses audience skepticism about product promotion. - [00:14:21](https://www.youtube.com/watch?v=V0YhpeSOuzk&t=861s) **From Idea to Intent: Hey Presto** - The speaker outlines the need for a simple, community‑driven tool to transform casual ideas into precise AI prompts, describes building Hey Presto to fill that gap, and invites feedback and collaboration. ## Full Transcript
Prompting is really weird because it's
the most wild west software space I've
ever seen, but it's also a highly
leveraged critical part of AI workflows.
There are dozens and dozens of prompt
tools. And as far as I can tell, very
few people have laid out or thought
through the overall life cycle of a
prompt and how we think about prompting
systematically. That is what this video
is for. I'm going to lay out how I think
about prompting in a life cycle. And I'm
going to lay out the tools that I think
are relevant at each stage. So, let's
jump into it. Let's start where most
people begin. Authoring and drafting.
So, you're writing, you're rewriting,
you're testing prompt text. I'll be
honest with you, a lot of people are in
Claude or in Chad GPT for this, right?
And they're like, "Hey, make it better.
Hey, make it better." Or people are
using tools like Prompt Perfect. Some
people are in cursor if they're in code.
This is all about hands-on
experimentation and wording refinement.
We're not testing the value of the
prompt. We're just trying to figure out
like there's a mental model we have of
the perfect prompt. And whether it's
right or not, we're trying to make the
prompt that we write fit that model. And
that's that authoring and drafting
stage. And what we find in this
situation is that the LLMs help us to
organize our thoughts and help us to
take these messy thoughts and kind of
clean them up. The next stage in
prompting is really around getting
serious about versioning. And so teams
will start to or individuals will start
to say this is a prompt I use often. How
do I keep it track of it when it needs
to change? And so teams will store their
prompts. They'll name them v1 v1.1.
They'll describe differentiation with
them. They'll diff them. They'll name
them. And prompts are then almost
treated like code. they become artifacts
in the business because they get reused
so much. Prompt layer definitely offers
tooling capability here. Prompt methus
does this. There are git based
approaches that do this. There are a lot
of other approaches as well. I know that
lang does this. The idea is you are
trying to solve for the persistence of
the prompt. So there's one record for
it. You need to make sure it's auditable
and you make need to make sure that you
can enable team level coordination on
the prompt. The next stage after
versioning is making sure that the
prompt is regularly tested. And this is
especially true if it's a production
grade prompt that will be used for an
LLM production system. And so you need a
situ a solution where you can compare
multiple prompts. You can evaluate
outputs for accuracy. You can evaluate
cost and hallucinations etc. Tools for
this include Hegel's prompt tools prompt
flow eval components prompt methus.
Again there's some custom eval
frameworks. There's a lot of other tools
around eval. A lot of people are writing
their own eval solutions and finding
that those are more flexible and more
effective because they allow them to
write to the detailed setups they have.
One of the places where we start to
differentiate between individual
builders and teams is right here at
evaluation and testing. Because before
if you were a serious individual, you
might still have your own versioning
system. You might still have a little
notion database where you keep track of
your individual prompts and it works for
you. When it gets to eval though, teams
who are using production grade prompts
will build entire suites of tests, 50
tests, 100 tests that they run in an
automated fashion in a pipeline against
a new version of a prompt. Whereas
individuals, we're very unlikely to do
that. we will probably test in an ad hoc
way or if we are super super organized
we'll have a library of 10 or 15 queries
that we are running with the prompt to
see if it works better but it's rare to
get super serious and so as we move
forward in this flow we see that these
prompt tools start to become more team
oriented this is one of the things that
makes it really complicated when you're
doing prompting because prompting is
both an individual productivity choice
and also something that supports teams
and dealing with that makes it hard to
write good software. But we're not done
with evaluation. After evaluation comes
constructing workflows or automation.
And this is when prompts become mere
steps and workflows. Like the prompt
becomes a guidepost for an agent that
might have tools, that might have memory
that it calls, that might have
conditional logic. And so then you begin
to need to build like with Google's
agent kit, with Langchain, with
Langmith, with Hegel and Prompt tools,
with prompt flow, with React agent for
frameworks. This is all about multi-step
automation and agentic behavior and
prompting. Again, like this is part of
why it's a tricky subject. It bleeds
into this area because really, if you're
trying to build an agent, you have to
consider the prompt as the beating heart
of the agent. The prompt is what helps
you to predictably guide the agent.
Finally, you have deployment tools. So,
prompts are embedded in real
applications and they need to be
tracked. They need to be up all the
time. They need to run correctly. Prompt
layer comes back through here. Langmith
comes through here. There are model APIs
that are directly available from OpenAI
and Enthropic. You have to have
production robustness, safety,
traceability, governance. You can see as
we've gone through this overall flow
that we have gone heavily into the teams
and companies world where prompting is a
piece of code that companies need to
maintain and they need production
tooling for that system. I am here to
tell you that you are missing a stage in
the prompting layer. Let's go back
through. What if we thought of our first
piece, authoring and drafting as stage
two, not stage one. Because it is.
Because when you think about where you
want to go with prompting, it's actually
intent formation and discovery that has
to happen first and then you get into
authoring. And what I have found sitting
down with people over the last few
months is that intent formation for
individuals is really hard. And this,
I'll be honest with you, this is
definitely something that is more true
the earlier you are in your AI journey
and you are still trying to figure out
how how prompting actually works. But it
remains true the farther you go in your
journey if you need help to trade time
for expertise. Basically, if you're
trying to write a prompt quickly and
formulate your intent quickly and you
don't have the time to do it and you may
be an advanced prompter, you still run
into the same issue because you have a
fuzzy goal. You summarize this, draft a
plan, analyze a sentiment, and now you
need to get to a structured,
unambiguous, high lever prompt that is
going to clarify the objective and the
constraints and the steps. And you know
that you should, right? Like we know we
should. It's like eat your vegetables.
There are not great tools at this stage
and most people use chat GPT. I'll be
really honest with you, most people use
chat GPT for this or they use Cloud for
this and that's fine. But what I have
found is that those tools by themselves
aren't super well suited to the kind of
intent formation that we're doing at
this stage. And there's a really really
simple reason why. When you're crafting
a prompt in Claude or Chad GPT or
Gemini, you are crafting it and
implicitly you are assuming the prompt
will work in that particular LLM.
there's not a cross LLM compatibility
check going on there. You are also
typically not getting the LLM to help
you think through what is the output
format for this prompt and artifact that
you're building. And I that that sounds
really abstract. So let me make it
concrete for you. If you have a fuzzy
intent, you typically know the output
needs to be a deck at the end of this
process. But what you don't often do is
you don't often say, "Please tune this
prompt in such a way that it's specific
for writing a deck because you're at the
fuzzy stage. You're trying to think
through the content first." And content
comes before format. What I'm picturing
for you is the realworld complexity that
I feel that others feel when they're
trying to craft prompts. And to be
honest with you, there hasn't been a
great tool for that. And that's why I
built one. I built Hey Presto to solve
for specifically this ideation intent
formation piece. And if you come back to
me and you say, "Nate, I'm using Chad
GPT." I will say, "Bless you. Have a
wonderful time. This tool is not for
everyone." Let me show you how it works
and then we'll get into the rest of it.
All right. Here we are in the prompt tab
on Hey Presto. And I have just pasted in
some very rough notes for an app I keep
trying to build called my family travel
app where it shows me the different
destinations around the world. and I
say, "Please help me craft a prompt to
build code for my app." All I do is I
say, "I want to build code." And it
gives me an expanded prompt. Look at
that. It's going to give me lots and
lots and lots of detail on the travel
app. It's going to give me a suggested
file structure, suggested data model.
And keep in mind, all of this is
editable. I have made this is ideation
stage, right? I have made no
commitments. Nothing is running. I can
go back through and edit all of this. I
don't have to believe or buy any of
this. And I can also change it. And so
if I want to change the stack, this is a
Python stack for those of you who are
engineers. If I want to change it and
say, look, I'm not writing this in
Flask. I'm writing this in React. That
makes more sense. All I have to do is
append a uh write this in React uh and I
can just regenerate and it will just
regenerate the uh regenerate the prompt
and it runs pretty quickly. While we're
uh letting it show up, you have
different tones you can get with this.
You can select different paragraphs or
options. JSON, email, table, numbered
list, step by step, structured sections.
Oh, it generated. Great. Um, and look at
that. It's getting into React. It's
getting into Tailwind, and it just
rewrites it. And so, it takes the detail
you give it, and in this case, I gave it
a fair bit, and it will turn it in to a
high-grade prompt. And if you're
wondering, does this actually work? I
will show you what I've been working on
in lovable. The answer is yes. So, here
we are in Lovable. We're crafting a cute
little app. We can add our destination.
It's not done yet. We can add a family
member. Little energy levels here for
the family member, whether they're an
adventure, relaxer. The app isn't done
yet, but it's coming together. And I was
able to use the prompt to craft the
initial piece of the app. And this is
not just for lovable, right? Like, this
is something I could have put into
anti-gravity from Google. It's something
I could have used uh for any other
vibecoded application or I could even
even have used it in cursor or other
spaces to get started. It is designed to
be agnostic of the tool but focused on
the outcome you want which I think has
been a space that's been missing
strategically for us because when we're
forming intent when we're at that fuzzy
initial stage that I'm arguing we don't
have good words for we need the
flexibility to go anywhere. Let me give
you one more example. This one's around
PowerPoint text. So, here we are with a
PowerPoint deck summary of Andre
Carpathy's famous early 2025 software
3.0 talk. It's gorgeous. It looks great.
Software is changing again. I just love
the look of this. All generated by one
prompt from Hey Presto. Super easy. And
I got oh, I don't know, 15 16 I got 20
slides out of it. Right. Embrace the
Iron Man suit. I love this. What a
finish. It makes sense. It's ready to
It's ready to share.
So, how did I do this? Very simple. All
I did was I said, I want to make a deck
about this article. I put in my notes.
Here's my notes on Andre's article. Dum
dum dum. I said, I want it to be a deck.
And I got create a deck, right? It has
everything. It's laid out. It's got the
slides laid out. It's all perfect. And
you can change this, right? You can
adjust it. And by the way, if you're
wondering, why would I go to Hey presto?
That's sort of annoying. Uh, I get it.
And maybe you don't want to. But if you
do want to, we've made buttons so it's
really easy to go into Claude or to go
into chat GPT directly. And so you can
open it up and it will just pop open the
prompt right in Claude. Here's the
prompt opened up right in Claude. Super
super easy. Now, this is the part where
people usually roll their eyes because
they're like, "Oh, Nate has sold out.
He's trying to like shill and like push
his stupid little product on us." And
I'll be honest with you, I could be the
person who does that and I'm not going
to because I'm not interested in saying
this is the best product for everything.
That is why I spent 15 minutes going
through all the stages of the prompt
tool chain and talking about all the
other great tools that you have to go
after for prompting. I do not believe in
a world where there is one prompt tool
for everything. And that in turn drives
the way I'm thinking about pricing. If
we are thinking about pricing for prompt
tooling, it needs to change depending on
whether you're a team solution or an
individual solution. And in my case, I
want to make this something that is very
very affordable if you're already a part
of the Nate Substack community. I don't
want this to feel extra fancy or
complicated. And so the easiest way I
can figure it out is to just say if you
are a member of the Substack, you will
get a coupon code for 70% off forever on
the prompt tool. which makes it like
seven bucks a month or something. And
I'm doing that because I want it to be
super easy for you to use. And if it's
useful, great. And fantastic, right? And
on my side, I'm committing to continuing
to make it better. And so, I'm going to
be opening up a Slack channel into my
working Slack for people who are using
Hey Presto so that they can give direct
feedback to me and a couple of other
builders who are working on this so that
we could make it better. My goal is to
solve that fuzzy intent stage. So if you
remember, we have these multiple stages
in the prompt tool chain. And I'm going
to show you what this looks like using a
slide created by Nano Banana. So here we
are. These are the six stages of the
prompt life cycle. Right? We have spent
almost all of our tooling and thinking
time around stage two and stage three.
Writing, rewriting, testing prompt text.
Stage three, storage and versioning.
This is where individuals are really
seeing prompt tools. And then there's a
lot of team level prompt tools as we get
into these more purple stages, right?
Evaluations, workflow construction,
deployment, and production integration.
I I really think this is important. If
you want to get from casual ideiation to
like clean intent, there needs to be a
good tool for that. And I just haven't
found a tool that more easily helps me
to name this is what I want to build.
It's a deck. It's a doc. It's
communications. You can do Slack
communications. It's code. Whatever it
is. And that's why I built Hey Presto.
If it's not for you, if you want to hack
around and chat GPT, bless you. It may
work well. I've certainly done it
before. And if you disagree with me and
you think intent formulation is not
correct for prompting, I would love to
hear what you think the correct initial
stages because I do not believe the
correct initial stage is writing or
authoring. Either way, I had fun making
Hey Presto. I think a lot of the future
of tooling and tooling development is
talking with communities, hearing what's
going on, and trying something. And I'm
offering Hey Presto in that spirit. If
you're part of my Substack community,
this is going to feel really affordable.
And it's going to be something that I'm
going to be plugged into on Slack so
that I can evolve it with you. My goal
is to make it useful. I don't need to
make it super big. I just want to serve
the community around a need that I
found. I also think it's good to
actually build things and launch them if
you talk about AI all the time. And so
that's another reason I'm putting this
out there. I think it's important to
have AI tools that solve problems if I
talk about AI tools a lot. And so that's
part of the reasons why I've been
involved in building this and digging in
and constructing this. If you guys are
interested, I'm happy to do a little bit
of a deeper dive on how I built the
tool. I wanted to start with a prompting
structure first because I think that we
have missed a framework for prompting
and I felt poorer until I could name the
stages. I felt like I had trouble
understanding my own thinking until I
could name the different stages of the
prompt tool chain and talk about the
tools because this is a world where like
no shade on all these other tools.
They're great. That's part of why I took
the time to introduce them. Like I think
that you need multiple tools for these
different stages and my goal is just to
help you with that intent and initial
sort of piece that I think hasn't been
as clean as it needs to be. So, hope you
enjoyed it. If nothing else, now you
have some vocabulary to talk about the
stages of the prompt and I think that
will be helpful and maybe you got some
good tool tips for some of the other
later stages. That's another reason I
made this video. I think there's a lot
of great prompting tools out there.
Cheers.