# Enterprise AI Pricing: Why Everything You Know Is Wrong

## Executive summary

The video argues that current enterprise AI pricing models (per seat or per token) are fundamentally flawed because they treat AI as a mere software purchase rather than a new economic input: intelligence. True ROI is achieved not by constraining spend, but by developing the internal capability—the 'manager of AI'—to allocate intelligence and integrate it into core business processes. The most valuable models involve deep partnerships and 'forward-deployed engineering' that focus on measurable business outcomes rather than usage metrics.

## Key takeaways

- AI is a New Economic Input: Historically, economies relied on people and capital. AI introduces 'intelligence' as a third economic input. The focus must shift from how to pay for the capability to how to allocate it to generate value.
- The Shift to Outcome Pricing: The ideal model is outcome pricing, where payment is tied to the delivered business result (e.g., a paid bill handled by an AI agent, as demonstrated by the Cedar example). This requires a deep partnership commitment.
- The Role of the 'Manager of AI': Individuals and teams must transition from being mere users of AI to 'managers of AI.' This involves making critical decisions about where to apply intelligence, what work deserves investigation, and when to transition back to human intervention.
- Forward-Deployed Engineering: This represents a key pathway for value capture, where engineers work directly inside or alongside a customer's organization to design, build, test, and deploy systems connected to the customer's data and processes, turning purchased intelligence into functioning systems.

## Technical details

- AI Pricing Models: The speaker critiques traditional models: 'Per seat' (traditional SaaS measure) and 'Per token' (a way to meter compute). While token pricing is useful for budgeting and experimentation, it does not guarantee a fixed amount of useful thinking, leaving the burden of value capture on the buyer.
- AI Agent Functionality: AI agents, such as 'Cora' (used by Cedar), are capable of handling complex, multi-step customer interactions (e.g., explaining a medical bill, confirming insurance, and processing payment) autonomously, demonstrating a closed-loop service interaction.
- Value Generation Process: Effective AI deployment requires a process where frontline engineers identify recurring failures, use AI agents to pull together maintenance records and compare cases, and then manage the intelligence—deciding when to use the AI and when to engage a human expert.

## Practical implications

- For organizations: Shift the conversation with AI vendors from cost reduction (cutting costs) to value creation (freeing up upside).
- For leadership/CFOs: When evaluating AI investments, put the company's ability to create value on the table alongside the ordinary pricing conversation.
- For employees: Focus on developing skills to become 'managers of AI' by proactively identifying process gaps and applying intelligence to generate measurable revenue or efficiency gains.
- For vendors: Consider deep partnerships or 'forward-deployed engineering' models that embed expertise within the client's operations to ensure value realization beyond simple usage billing.

## Topics

AI Economics, Enterprise AI, Value Chain Management, Outcome Pricing, AI Strategy, Process Automation, Shopify Q2 2026 Financial Results, OpenAI's deployment company

Source: https://www.youtube.com/watch?v=RvnQZQAHUDc