Models Are Commodities Now
The race moved to application. The model layer is turning into infrastructure. What sits on top of it, close to a specific user and workflow, is the new differentiator.
Mike Molinet & Govind Kavaturi
Last week we talked about engineers got AI that works. Everyone else got a chat box. Same model, different outcomes. The model isn't the gap. The setup is.
This week: the race just moved.
The Signal
Monogram launched today. $40M seed.
Not a model company. An application company.
They're not selling intelligence. They're selling what you do with it. How it fits into your workflow. How it feels to use.
Nobody reading about Monogram asked what model powers it. Nobody cared.
That's the signal.
Models Are Commoditizing
A year ago, the race was: who has the best model?
GPT-4 vs Claude 3 vs Gemini. Benchmarks. Context windows. Reasoning scores.
That race is flattening.
Everyone gets the same model now. The upgrade won't save you. Same intelligence, same API, same capabilities.
The model layer is becoming infrastructure. Like AWS. Like Stripe. Essential but invisible.
Nobody asks what cloud provider powers their favorite app. Soon nobody will ask what model powers it either.
Value Is Shifting
If the model is a commodity, where does value go?
Application.
UI/UX. Ease of use. How you bring AI to a specific use case. How deeply you understand the workflow. How seamlessly it fits into someone's world.
Stop selling AI. Sell the outcome. The application IS the outcome.
The model is what makes it possible. The application is what makes it matter.
The Harness Is The Product
This is why harness engineering is becoming the discipline.
The harness is everything around the model. Prompt architecture. Memory systems. Tool integrations. Orchestration logic. Context management. Output handling.
None of this appears in benchmark scores. All of it determines whether the thing actually works.
The teams winning with AI in 2026 aren't the ones with access to the best model. They're the ones who've built the best harness around a capable one.
The harness is the product. The model is interchangeable.
Why This Is Great News
If you're a founder or entrepreneur, this is the opening.
You don't need to build the model. You don't need billions in compute. You don't need a research lab.
You need to understand a use case deeply. You need to apply the model better than anyone else. You need to build the thing that makes AI actually work for a specific person with a specific problem.
The hard part isn't intelligence anymore. The hard part is application.
And application is something founders can do.
Claude And OpenAI Can't Please Everyone
Anthropic and OpenAI will keep building general consumer products. Chat interfaces. Assistants. Tools for everyone.
That will only go so far.
AI is deep work. It needs deep understanding of consumer behavior. Specific workflows. Specific pain points. Specific contexts.
A general assistant can help with a lot. But it can't replace the thing built exactly for your problem.
Build something people want. The application layer is where "want" lives. Where you understand urgency. Where you predict behavior. Where you know the user journey better than anyone.
The foundation model companies can't go that deep for every use case. That's where you win.
The New Stack
The stack is inverting.
Old stack: Model is the differentiator. Application is a wrapper.
New stack: Model is infrastructure. Application is the differentiator.
The model provides intelligence. The application provides value.
The model is general. The application is specific.
The model is a commodity. The application is a moat.
What This Means For Builders
Stop asking "what model should I use?"
They're all good enough. Pick one. Move on. The model choice is not your edge.
Start asking "how deeply do I understand this use case?"
Who is the user? What's their workflow? Where do they get stuck? What would make them pay? What would make them stay?
Engineers got AI that works because someone built Claude Code and Codex specifically for how developers work. Not a general chat. A specific application.
That's the template. Who else needs their version of Claude Code?
Build for the 90%.
The other ninety percent are still stuck with chat boxes. They're waiting for someone to build the application that makes AI work for their job, their workflow, their problem.
That's the opportunity.
Monogram Is Not The Exception
Monogram raised $40M because investors see it too.
The model race is over. The application race is starting.
Expect more of this. Application companies raising serious money. Not because they have a better model. Because they have a better understanding of a specific user.
The capital is moving. The opportunity is moving.
Are you?
The Close
Models are commodities now.
Same intelligence. Same APIs. Same capabilities.
The race moved to application.
Who understands the use case? Who builds the harness that makes it work? Who knows the user deeply enough to make AI actually useful for them?
That's the game now.
The model is the floor. Application is the edge.
Nobody cares what model you use. They care what it does for them.
Vol 1: The opportunity exists. Economics changed.
Vol 2: Smart operators doing dumb work. Patterns to spot.
...
Vol 23: Engineers got AI that works. Everyone else got a chat box.
Vol 24: Models are commodities now. The race moved to application.
The model is the floor. Application is the edge.
Mike & Govind
Look at your product. If you're leading with the model, rewrite it. Lead with what it does. For whom. In what workflow. The moat is not what powers it. The moat is how deeply you know the person on the other side.