The End of Token-Based Productivity

The era of measuring AI productivity by the volume of output is over. Eric Siu and Neil Patel note that the industry is moving toward outcome-based pricing, where the value of an AI agent is tied directly to business KPIs instead of computational spend. This transition forces a change in human labor: we are moving away from prompting toward a role Siu calls the forward-deployed marketer. This role requires high-level strategy and system orchestration, as the actual labor is handled by agents. For organizations, this creates a competitive divide: those who treat AI as a tool for efficiency will be outpaced by those who use it to reinvent their business models. The advantage goes to those who prioritize compounding leverage over immediate, superficial output.

The Shift from Volume to Value

The most important takeaway from the conversation is that token usage is a poor way to measure productivity. When teams focus on token volume, they often generate polished but useless content. The system responds to this inefficiency with adaptive routing, a mechanism that balances quality and cost by selecting the right model for the task. This is more than a technical fix; it is a move toward accountability.

"If Devin AI delivers less engineering value than you are paying for, Cognition will fund your usage until it does."

-- Eric Siu

This guarantee changes how we evaluate technology. When the provider assumes the risk of the outcome, the focus shifts from how much we are using to what we actually achieved. This creates a downstream effect where engineering and marketing teams must connect AI agents directly to revenue-critical events, such as production incidents or stalled sales deals, rather than using them for theoretical tasks.

The Rise of the Forward-Deployed Marketer

As AI agents take on the burden of labor, the most valuable human hire is no longer the doer but the forward-deployed marketer. This role is defined by ownership of the client outcome, while the agents handle the execution. This is a difficult transition for most organizations because it requires a different set of skills: the ability to build, strategize, and orchestrate agent workflows rather than simply managing tasks.

The systems-level implication is that the marketing agency model is being forced to evolve. The traditional services model, which relies on billable hours for reporting and manual execution, is being replaced by AI. The forward-deployed marketer thrives by reducing client calls and reporting time, focusing instead on the high-leverage strategy that compounds over time.

"The one question that compounds your results over a decade: what is the highest-leverage thing I can do with my time today?"

-- Eric Siu

This question acts as a filter for all activity. For Neil Patel, this means prioritizing deal-making and partnerships over content creation, which his team already automates. The competitive advantage is found in the willingness to abandon outdated models, even when those models feel safe or familiar.

The Compounding Cost of Resistance

The conversation reveals a reality regarding the wealth gap in AI fluency. Because the benefits of AI-driven productivity compound, the distance between those who embrace the technology and those who resist it will grow. Resistance to change often comes from a desire to return to the way things were, a sentiment Siu and Patel identify as a primary failure mode for legacy businesses.

The system responds to this resistance by routing around it. As search queries evolve into long-tail, conversational questions, businesses that fail to adapt their content strategy to this new interface will become invisible. The highest-leverage work today involves reinventing the business model, like publishers selling data to LLMs or adopting paywalls, rather than fighting to preserve a dying status quo.

Key Action Items

• Audit your AI spend (Immediate): Stop measuring productivity by token usage. Shift your internal metrics to revenue-critical KPIs. If an agent is not moving a specific business metric, it is a liability, not an asset.
• Identify your forward-deployed talent (Next Quarter): Evaluate your team for individuals who can orchestrate agent workflows rather than just executing tasks. If you are not staffing for this, you are already behind.
• Implement the highest-leverage check (Daily): Start every morning by asking: "What is the single most important thing I can do today that will compound over the next decade?" Use this to cut low-leverage tasks.
• Transition to outcome-based contracts (12-18 months): Seek vendors who provide AI productivity guarantees. If a vendor will not tie their pricing to your outcomes, they are not aligned with your success.
• Reinvent your business model (12-18 months): If your current business model, such as traditional journalism or manual reporting, is being disrupted by conversational search, stop trying to preserve the old model. Sell your data, implement paywalls, or shift to high-value, proprietary content.