AI agent-based pricing: will it become a reality?

2026-07-23 · Source: Everest Group Research Portal · Field: Business & Management — Operations & Process Management, Consulting & Professional Services, Corporate Strategy & Leadership · Depth: Intermediate, short

Summary

AI agent-based pricing, where an AI agent is a countable commercial unit, is a widely discussed concept in the IT-BPS outsourcing ecosystem. While currently observed in some deals, its mainstream adoption faces significant challenges. Unlike human agents, which are measurable units delivering approximately 1,800 productive hours annually from offshore locations, AI agents lack consistent properties for commercial measurement. The number of AI agents deployed depends on architectural and design choices, not the actual quantum of work, making it disconnected from both provider costs (inference, compute, infrastructure) and client value (STP rates, cost takeout, DSO). The article highlights that agent count is an implementation detail, not a meaningful commercial metric, and raises questions about versioning and billing for enhanced agents. Instead, current successful deals utilize outcome-based or output-based pricing models, with AI tool run costs integrated and development fees amortized.

Key takeaway

For CTOs and procurement leaders evaluating BPO contracts involving AI, avoid adopting AI agent count as a primary commercial metric. Your focus should remain on outcome-based or output-based pricing models, such as per invoice or gainsharing from improved DSO, which directly align with business value. Ensure development and deployment costs for AI solutions are amortized or treated as provider investments, rather than being tied to an arbitrary agent count that doesn't reflect underlying costs or value.

Key insights

AI agent count is an implementation choice, not a viable commercial pricing unit for outsourcing.

Principles

• Human agent pricing relies on countability, boundedness, and linear workload-revenue.
• AI agent count is a function of architecture, not work quantum or client value.
• Provider costs and customer value are decoupled from AI agent count.

In practice

• Price outsourcing deals based on outcomes (e.g., DSO improvement, cost takeout).
• Price outsourcing deals based on outputs (e.g., per invoice, per conversation).
• Amortize AI solution development and deployment charges over the deal term.

Topics

• AI Agent Pricing
• BPO Outsourcing
• Outcome-Based Pricing
• Output-Based Pricing
• Commercial Models
• AI Inference Costs

Best for: VP of Engineering/Data, Director of AI/ML, AI Product Manager, Consultant, Executive, CTO

Related on AIssential

• AI Agents Are Breaking Traditional SaaS Pricing — AI agents shift SaaS value from human access to completed work, necessitating new hybrid pricing models.
• Consultants Are Cashing in on the AI Boom — AI is proving to be a valuable tool for efficiency and expansion in both journalism and consulting, rather than solely a disruptive force.
• Salesforce’s Outcome-Based Pricing for Agentforce Help Agent: What Enterprise Buyers Need to Negotiate Now — Salesforce's outcome-based pricing for Agentforce Help Agent requires diligent negotiation within the broader vendor relationship.
• [AINews] Codex Rises, Claude Meters Programmatic Usage — AI competition intensifies with strategic pricing, advanced agent infrastructure, and significant research breakthroughs.
• Your service vendors are being rebuilt around AI — AI-driven service vendor consolidation and outcome-based pricing demand rigorous buyer-side governance to mitigate significant risks.
• Don’t Just Talk About AI. Measure Business Outputs. Here’s How. — Measure AI success by business outputs, not activities, to ensure tangible value and avoid pilot failures.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Everest Group Research Portal.