AI Musings - Pricing Dilemma. AI has made it cheaper than ever to build, but expensive to run — exposing a gap between how value is delivered and how it’s priced. While outcome-based pricing is becoming the ideal, most still default to time or feature-based models because true impact is hard to quantify. The real challenge is figuring out how to price cognitive leverage — not just effort. This post explores that tension and invites others to share how they’re navigating pricing in an AI-driven world.

We Know How to Build AI — But Not How to Price It, just yet.

If the value is in the outcome, why are we still pricing the hours?

That’s what I’ve been reflecting on since last week’s Tech Alpharetta Startup Circle.

A patent law consultant said: “We price for time because outcomes are unpredictable.” He wasn’t wrong. In his world, results could take years—or never materialize.

But here’s the scenario I keep running into:

→ In AI-driven work, outcomes are the pitch. → But in real-world projects, effort is still the billing unit.

We’re living in that messy middle — where pricing models haven’t caught up with the tech they’re meant to support.

Yes, it’s cheaper than ever to build. But not to run.

You can launch a GenAI MVP with “vibe coding bros” for under $1,000. But run it with real users for 3 months?

Suddenly you’re looking at: • $4K+ LLM usage costs • Bloated Vector DB costs • An orchestration stack that’s more duct tape than design (thanks to “workflow bros”) • And rising cloud bills you didn’t plan for

It’s the “printer is cheap, cartridge is expensive” trap.

Consultants are told to price outcomes. Startups are told to price value. But most of us still default to hours, features, or usage tiers.

Why?

Because we still don’t know how to price cognitive leverage— The real output of AI systems.

So here’s what I’m exploring: • What’s the right unit of value in an AI engagement? → Time saved? Tasks replaced? Better decisions? • Should products price per result, not just per token? • If an agent helps improve a $10K decision — should that be a 1% fee?

Problem → We price inputs because outcomes feel fuzzy Opportunity → Price leverage — not just labor

Maybe a new middle ground will emerge: • Cost transparency like a utility bill • Value alignment like a success fee

We’ve figured out how to build intelligence. Now we need to price for impact.

Until then, we’re all experimenting.

Curious to hear how you’re approaching pricing: Projects? Products?

1

2

3