Hi Reader

The second episode of Intelligent Artifice just dropped: and this one’s a masterclass in building AI-first before it was a trend.

I sat down with Andrew Skrypnyk, CEO of Promova: one of the most prolific advertisers on Meta, and one of the earliest to integrate AI across their entire product and operations stack.

What’s wild is:

They were building AI-driven features long before GPT hit the headlines.

Not with prompts: but with custom machine learning models.

In this episode, Andrew and I get into:

✅ How Promova built pronunciation models before LLMs existed

✅ Why LLMs still can’t do what ML models do best (like audio analysis)

✅ How they automated customer support: and increased user satisfaction

✅ The exact use cases where they still rely on hand-written code

✅ Why AI tools work best as mirrors, not masterminds

✅ How we’re entering the App Store era for agents

….and a TON more.

It’s a founder who’s built AI into every corner of his company — and is doing it at scale.

🎧 Check it out now.

(Links to all platforms + full transcript included.)

Coming up next on the podcast…

✅ A team that scaled from image model adoption before it was cool

✅ A founder who shifted from ML to LLMs without ever saying “AI”

✅ A team that 10×’d creative output by redesigning their workflow

✅ A company doing millions in revenue per employee using AI

✅ The most surprising ways consumer subs apps are using AI

If you’re building, testing, or just trying to stay ahead in an AI-first world…

I think you’ll find these conversations worth your time.

Let me know what you think when you check them out.

Later,

Shamanth

​