AI cost management: How we track and optimize AI spend at Zapier

Zapier’s COO describes AI cost management as a way to ensure AI investments generate returns beyond their direct costs by tracking both subscription seats and token-based usage, then matching tools and models to appropriate tasks. After initially providing broad AI access with limited measurement, Zapier built an automated system that collects daily usage data from tools including Anthropic, Cursor, OpenAI, and its Sidekick agent harness; assigns individualized monthly allocations; alerts employees at 75% of their limits; and automatically evaluates brief budget-increase requests while escalating repeated requests to managers or moving employees into higher tiers. Employees also receive personal spend reports by tool and model, helping them make informed choices without restricting normal AI use. The approach emphasizes visibility, role-based limits, regular reviews tied to business outcomes such as revenue, productivity, cost avoidance, speed, and risk, and education about using lower-cost models or conventional deterministic automation for predictable work. Zapier also promotes “deterministic AI,” in which fixed rules handle repeatable workflow steps and AI is used only where judgment is required, with model selection managed centrally to reduce unnecessary token spending.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.