Link

Article: Your “Per-Seat” Margin is My Opportunity

What the article covers

Two unstoppable forces are colliding as we head into 2025: LLMs are making agents genuinely useful, and every corporation is demanding efficiency.

Per-seat pricing only works when your users are human…As AI gets more reliable, we’ll shift to pure outcome-based pricing.

Executives aren’t evaluating software against software anymore. They’re comparing the combined costs of software licenses plus labor against pure outcome-based solutions.

The winning strategy in my books? Give the platform away for free. Let your agents read and write to existing systems through unstructured data—emails, calls, documents. Once you handle enough workflows, you become the new system of record

Zero upfront costs, pay only for outcomes—that’s not just a pricing model. That’s the future of business.

My Thoughts

Overall takeaway

Charging for AI Agents is all over the place.

I read through 100+ AI Agent company’s pricing pages to see how they were structured.

Most still have the SaaS model of small team, medium team, large team, and Enterprise/Call Us.

A few have usage based pricing.

The surprising thing to me was how low the prices for many of these services given these companies really are aiming at reducing / replacing human labor.

In some sense, this is the story of technology being deflationary.

In another sense, this is also a pricing tactic to make it easy to adopt the new technology and grow quickly from there.

Given that Nikunj Kothari is a Venture Capitalist at storied firm Khosla Ventures, I’m very intrigued at the idea of giving away the platform for free.

Certainly places like Microsoft and Google may eventually start offering AI Agents for free in the hopes of charging somewhere else.

This makes the current window particularly interesting for startups who can move faster than the tech giants.

Strategic Implications

If we take this as true, then it means

• LLM inference costs must be eaten by the company
• Outcomes value must be high enough to pay per outcome
• AI Agent companies must be able to calculate this cost to breakeven

Additionally, to build an AI Agent company, you’ll need:

• Server to use the agents programmatically
• A tool execution sandbox so that AI Agent tools don’t interfere with the main server process
• API designs to enable multi user and multi agent support
• Database to persist agents
• ORM to scale the agents
• LLM (local or external) and the code for it
• File storage to do Rag if necessary
• Context Management systems to handle long term meory and potential context overflow

While many of the above can be solved by a AI Agent Framework and Build Your Own AI Agent Websites, you still have to pay for those services before you can deliver the outcome the organization will be willing to pay for.

Market Analysis

The current AI Agent pricing landscape reveals three patterns:

1. Traditional SaaS tiers (most common)
2. Usage-based pricing (emerging)
3. Outcome-based pricing (frontier)

The trend toward lower initial pricing suggests:

• Companies prioritizing adoption over immediate revenue
• Recognition of technology’s deflationary nature
• Strategic positioning for future value capture

When the top 20% of SaaS companies have gross margins above 80%, there’s significant room for AI Agent companies to undercut pricing while maintaining profitability through outcome-based models.

Jeff Bezos famously said “Your margin is my opportunity.”

This suggests we’ll see a wave of AI Agent companies targeting high-margin SaaS products, offering similar functionality but with outcome-based pricing.

The key differentiator won’t be features, but the business model itself.

Economic Model Considerations

The challenge lies in balancing two sets of needs: the customer’s need for clear value measurement and the AI Agent company’s need for sustainable economics.

Success requires satisfying both simultaneously.

For outcome-based pricing to work, companies (both AI Agent company and the organization they serve) need:

1. Clear Definition of Outcomes
   • Measurable success metrics
   • Agreed-upon validation methods
   • Clear value attribution

For outcome-based pricing to work, the AI Agent company has to focus on

1. Cost Management
   • LLM inference costs
   • Infrastructure overhead
   • Development and maintenance costs

and

2. Value Pricing
   • Understanding customer’s alternative costs (combined costs of software licenses plus labor)
   • Measuring and communicating ROI
   • Building pricing power through data accumulation

Implementation Framework

The window for this opportunity may be limited - as traditional SaaS companies adapt their pricing models, the advantage of being first with outcome-based pricing will diminish.

The key is moving quickly while maintaining reliability

Companies pursuing this AI Agent model should:

1. Start with Clear Use Cases

   • Choose workflows with measurable outcomes
   • Focus on high-value, repeatable tasks
   • Build clear success metrics

2. Develop Infrastructure

   • Robust monitoring and measurement systems
   • Scalable processing architecture
   • Clear data ownership and privacy controls

3. Build Trust Progressively

   • Start with parallel runs alongside existing systems
   • Demonstrate clear value metrics
   • Gradually increase autonomy and scope

And at all times focus on delivering (and owning) the full outcome.

Looking ahead, successful AI Agent companies will likely be those that:

• Move quickly to establish outcome-based pricing models
• Build robust infrastructure to deliver and measure results
• Focus on high-margin opportunities where the value proposition is clear
• Maintain ownership of the full workflow while integrating with existing systems

The opportunity extends beyond building better AI Agents toward fundamentally reshaping how businesses create and capture value.