Apr 2 (edited) • General discussion
🤖 My "Pre-Sell" Method of Building A.I. Tools That Work
Building with AI costing an arm & leg? This is how I use Travis's pre-sell methods to save time, energy, and money when building A.I. tools that work.
What's up peeps!
I was having a chat with someone this morning who's been running into coding costs with Claude Code.
Them tokens go REAL quick if you're not careful.
And the money it costs is REAL.
Now, I can barely spell coding.
But I've built enough tools at this point...
...that I've "scrabbled" smart ways of saving time & money.
I shared with him how I start with Claude CoWork first and THEN build out using Claude Code.
But the truth is I start even BEFORE that with Claude Chat.
This might save you a ton of time, energy, and tokens...
AKA: cash, moola, or dineros.
Plus you'll be able to help people easier.
Let's dig in.
✅ #1: I start with Claude Chat
Just like in Make 'Em Beg To Buy...
Before diving into making an offer...
You wanna UNDERSTAND...
Shared with me early on, "If you wanna double your income double your understanding of your prospective buyers experience..."
Brainstorm the idea.
Tell Claude what OUTCOME you want.
Don't tell it what to "DO".
When you tell it what outcome you want...
Ask it what the best approach would be.
After it gives you it's answer?
Ask it for feedback.
~ What are the pros and cons of this approach?
~ What questions am I not asking that I should be asking?
~ What challenges are going to come up that I'm not anticipating?
~ How would you best solve these in the brainstorming phase?
~ Acting as a world class [insert expertise] what's the smartest way to approach building this in phases?
You'd be surprised, when you lead with the OUTCOME you want first and let Claude brainstorm approaches, just how efficient it can be.
✅ #2: Using Claude CoWork:
CoWork is where I start proofing each piece of my new solution.
Just like Travis taught me early on...
Instead of going from new idea to selling a big "group offer" thing for $2000 (without it being 100% proven and solid)...
...you work with peeps 1on1 FIRST to get it nailed down.
Get all the quirks worked out.
Make sure they get the result.
THEN you roll it out to a small group as a "done with you".
Make sure you get feedback.
Work out more quirks as you're teaching one to many.
NOW, once nearly everyone is getting the result...
...you can roll it out as a "do it yourself" offer.
The same thing is true when it comes to building A.I. tools.
I have Claude Chat architect for me what I'm thinking based on our brain storming chat.
Then I create each phase or piece of the tool in Claude CoWork.
I test each piece and make sure each phase works as intended.
Skills are the NEW prompt because they relieve me of having to constantly copy and paste prompts over and over.
Plus I can "chain" those skill files so they run in sequence to accomplish what I want.
However, I also get real world feedback when dialing in my new Skill.
Is it working like I want?
Again, I can use Claude CoWork to help me answer questions we didn't anticipate in the Chat phase.
Like when rolling out a new offer in the group phase, sometimes you gotta iterate.
But once that's dialed in.
NOW I can tell CoWork I want to build this in Claude Code and to fully architect it out for me based on what I know ALREADY WORKS.
It'll provide me with the skills and files I need to upload into Claude Code.
Now when I start coding it's pretty frictionless!
All because I know the individual parts work where I was able to isolate the variables.
Nerdy?
Maybe...
Saves me a ton of frustration and energy?
100%
If this was useful lemme know below?
I might share more if so.
In your corner,
~ Tony
19
13 comments
🤖 My "Pre-Sell" Method of Building A.I. Tools That Work
skool.com/royaltyrockstars
Turn your courses, content and coaching (life) experience into an investment portfolio that sends you royalty checks for life.
Leaderboard (30-day)
1
+2
2
+1
3
+1
4
+1
5
+1
Powered by