AI with ALLIE

The professional’s guide to quick AI bites for your personal life, work life, and beyond.

I want to show you my secrets for how I unlock richer ways of working with AI. This is how I give design feedback, how I make decisions faster with my team, and how I manage my agents. If you want to learn more about the AI agent management side, I want you to join me and Mark Cuban at my free AI agent workshop on August 11 - save your seat here.

I built a profitable app - here’s how you can do it too

As a non-engineer, I have already built a profitable app that tens of thousands of community members have taken (loving the positive feedback!).

I want to walk you through exactly how I used AI to supercharge that app development process. And the main trick might surprise you.

For those of you that haven’t seen it, I built out the AI-First Index, an AI benchmarking system based on my years of advising Fortune 500 companies on how to win with AI. I took my personal expertise, my product mindset, my previous client evaluations, and my AI know-how to build the entire thing - all landing pages, automations, authentications, scoring - by myself. And AI (both Claude Code and Codex) helped me bring it to life.

A couple of years ago, that work would have taken 3-4 dedicated people and tens of thousands of dollars.

My first hack: I dictate everything.

As I've said countless times before, dictation is not only four times faster than typing, it is also easier to get more context out. I’m just rambling! It’s natural! Sometimes I'll go on a 45-minute walk and just non-stop dictate to Otter AI or Fireflies AI. Sometimes I’m resting, pensive, at my standing desk, dictating one or two sentences at a time. My favorite is when I am in live voice mode, leaning back in my desk chair, just watching my monitors and laptop function as an AI command center.

The absolute best thing you can do for yourself is download Wispr Flow and start to dictate at least half of your Google Docs or emails or Slack messages, basically anywhere that you're typing today. I want you to try dictation.

My second hack: start small. 

In order to build out this whole app, I had to create:

• a master landing page

• individual test landing page

• company test landing page

• company registration page

• company testing page*

• company base results page

• company full results page

• company payment confirmation page

• payment confirmation email

• individual testing pages

• individual results page

• individual email

• all of the backend logic to actually score both quizzes

• all payment logic

• dictation capabilities

• guardrails (think: API limits, spend limits, prompt injection protections)

• plus, followup marketing emails (still working on this one!)

If you tried to do that in one go, you would curl into a ball and cry your face off. I'm serious. Not even an engineer with 20 years' experience would have taken all of that on in one go. So what did I do? I started with the page that I added an asterisk to above. That's it. Just the company testing page. I mostly wanted to see if something like this could work and if it was able to replicate the same sort of evaluations that I do with my clients.

I built out that one page using our Brand Guidelines Template. I argued with Claude back and forth on what I wanted the experience to look like and feel like. I changed my mind several times. And I got to something that was good enough to move on to the 20+ other tasks. It wasn't perfect, but it worked - and that’s the goal here.

My third hack: iteration locally.

I'm gonna keep this advice short. I just want you to test this on your computer so you can see what it looks like, see what it feels like. Don't test in a live environment. Use the right-hand side of Claude Code or Codex's user interface, and don't constantly deploy your changes to a public-facing website where your customers could be facing a broken feature or link.

When you've already launched your app and you really get into the swing of things and have real users, I recommend using GitHub to deploy something called “test branches” before pushing code to the main public branch. It’s like giving yourself a sandbox to make sure things don’t break in the real world. And if those words mean nothing to you, I got your back - I put together an entire tutorial on how to build apps using Claude Code, GitHub, and Vercel.

My fourth hack: dynamic web design.

Imagine having to give feedback on all of those deliverables above with just your words, something like “can you go ahead and take the third from the left yellow box, but not that box, just the other box, not the big box? If you could go with the small box, and then, if you could make that box just a little bit bigger and then delete the box to the right of it.” I would rather take a cold shower every night at 3am.

No! We have AI that can code us anything. So let's use it to our advantage and ask AI to code us a design feedback page. Let me show you what I mean.

My fifth hack: create on-demand feedback interfaces.

Similar to the design feedback interface above, you can actually apply that same method to really anything you want to give feedback on. I do this in about 30% of my conversations with AI. The ‘build-me-an-interface’ hack works on any task where visuals will get you further than paragraphs of text. So if you need to manage a to-do list, a home renovation plan, a complex decision, a kid’s calendar, a team’s project plan, an org chart redesign, a shopping list, a travel itinerary - lean into your lovely design prompting skills and feed your visual cortex.

Let me show you some examples - the first is an early draft from the same AI-First Index Report I covered above.

You can watch me build all of the examples above here.

Or try it yourself: pick the one task that has multiple decisions to make, or where you keep re-explaining yourself to Claude and getting nowhere, and paste in a prompt like this:

AI-built interfaces are also how I manage my multiple agents. I have a mission control dashboard so I can see what’s being worked on live, and an AI boardroom so I can see how they debate my strategic decisions.

If you want to learn more about how I manage my 34-agent workforce, and all the latest on AI agents and AI employees, join me and Mark Cuban on August 11 at 1pm ET as we host a free AI Agent Workshop.

Stay curious, stay informed,

Allie

FREE AI AGENT WORKSHOP WITH MARK CUBAN

On August 11 at 1pm ET I'm running a FREE workshop with Mark Cuban on AI agents and AI employees.

Mark and I will be discussing where the real business opportunities with AI are in 2026, and his clearest advice for self-starters today. I'll also walk you through my whole 34-agent AI workforce, from my first AI employee to the chief of staff that routes everything and the six directors under it.

This was the fastest registration rate we’ve ever had for an event. I don’t want you to miss it.

Everyone that registers will get the recording as a replay.

Feedback is a Gift

I would love to know what you thought of this newsletter and any feedback you have for me. Do you have a favorite part? Wish I would change something? Felt confused? Please reply and share your thoughts or just take the poll below so I can continue to improve and deliver value for you all.