Enterprises Build Infrastructure To Scale AI Agents

Anthropic published a breakdown this week of nearly one million production agent tool calls showing software engineering accounts for 49.7%, sales 4.3%, and finance 4.0%. The article argues low adoption outside coding reflects data-access, integration, and feedback-loop limitations rather than model capability, noting CRM vendors and Gartner forecasts predict rapid infrastructure-driven adoption by mid-to-late 2026. Enterprises should prioritize data integration, governance, and observability to scale agent deployments.
Key Points
- 1Shows 49.7% of production agent calls in software engineering; sales 4.3% and finance 4.0%.
- 2Explains coding and support advantage: centralized data and immediate, verifiable feedback loops accelerate agent scaling.
- 3Recommends investing in CRM-native integrations, governance, and observability to unlock sales and finance agents.
Scoring Rationale
Strong industry relevance and actionable guidance on integration infrastructure; slightly limited by being synthesis of existing vendor and analyst reports.
Sources
Public references used for this report.
Practice with real SaaS & B2B data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all SaaS & B2B problems
