You need a legal AI. You search for "legal AI." You find dozens of options. They are all fine-tuned for legal work. They know contracts. They know case law. They know legal writing. They are not generalists. They are specialists. This is the Model-as-a-Service (MaaS) business. Companies are selling fine-tuned models for narrow use cases. Vertical AI is exploding.
This is a fundamental shift. The market is moving from general-purpose models to specialized ones.
What Is Model-as-a-Service?
Model-as-a-Service is a business model for selling AI models.
The Concept:
A company fine-tunes a model for a specific use case.
They sell access to the model.
They charge per use or per subscription.
The Examples:
Legal AI: Harvey, Casetext, LawGeex.
Medical AI: DeepMind Health, IBM Watson Health.
Coding AI: GitHub Copilot, Cursor, Replit.
A Contrarian Take: MaaS Is Not New. It Is a Return.
We call it "new." But it is a return to the old model. In the 1990s, we bought specialized software for specific tasks.
MaaS is just specialized software for the AI era.
The Rise of Vertical AI
Vertical AI is growing rapidly.
The Drivers:
General-purpose models are not good enough for specialized tasks.
Fine-tuning improves performance.
Companies are willing to pay for specialized models.
The Result:
A proliferation of specialized models.
A fragmented market.
High prices for narrow use cases.
A Contrarian Take: Vertical AI Is a Bubble.
Vertical AI is a bubble. The market is too fragmented. The prices are too high.
The bubble will burst. The generalists will win.
The Economics of MaaS
MaaS is a lucrative business.
The Revenue:
Subscription fees.
Per-use fees.
Enterprise contracts.
The Cost:
Fine-tuning costs are low.
Inference costs are low.
Margins are high.
The Profit:
MaaS is highly profitable.
The companies are growing rapidly.
A Contrarian Take: The Economics Are Not Sustainable.
The economics are not sustainable. The competition is intense. The prices are falling.
The companies are in a race to the bottom.
The Generalist vs. Specialist Debate
The debate between generalists and specialists is ongoing.
The Generalist View:
General-purpose models will eventually be good enough.
They will subsume specialized models.
The specialists will be irrelevant.
The Specialist View:
Generalists will never be good enough.
Specialized models will always be needed.
The specialists will thrive.
A Contrarian Take: The Debate Is Misguided.
The debate is misguided. Generalists and specialists will coexist. They serve different needs.
The future is not one or the other. It is both.
The Future of MaaS
The future of MaaS is uncertain.
Near Term (1-3 Years):
Vertical AI will continue to grow.
More companies will enter the market.
Prices will fall.
Medium Term (3-7 Years):
Generalists will improve.
Specialists will adapt.
The market will consolidate.
Long Term (7-10 Years):
Generalists will dominate.
Specialists will be niche.
The market will stabilize.
A Contrarian Take: The Future Is Not Consolidation. It Is Fragmentation.
The future is not consolidation. It is fragmentation. The number of specialized models will continue to grow.
The market will become more fragmented, not less.
What This Means for You
You are a consumer of AI. You have choices.
- Choose the Right Model:
Generalists are good for broad tasks.
Specialists are good for narrow tasks.
- Consider the Cost:
Generalists are cheaper.
Specialists are more expensive.
- Be Aware of the Lock-In:
Specialists may lock you in.
Generalists are more flexible.
The Last Model
The last model is not a specialist. It is not a generalist. It is a choice.
You ask: "Should I use a generalist or a specialist?"
The AI says: "It depends."
You realize: The choice is not about the model. It is about the task.
If you could choose between a generalist model that is good at many things and a specialist model that is excellent at one thing, which would you choose? And why?
Top comments (0)