August 2026 Summaries

A Retool survey of 101 senior technology and security leaders at mid-market and enterprise companies found that 43% exceeded their 2026 AI budgets, although most organizations continue expanding AI adoption rather than reducing spending. More than half budgeted $100,000 to $1 million for AI tools, while 24% allocated over $1 million, and respondents cited limited visibility into production AI systems, growing token use from agentic workflows, and ungoverned “shadow AI” as central contributors to unexpected costs. If infrastructure costs rose 20%, leaders most often said they would consolidate vendors or audit low-impact usage rather than absorb the increase, reflecting a growing emphasis on model routing, token tracking, and usage controls. Vibe-coded software was reported as a moderate or major spending contributor by 52% of respondents, with potential costs extending beyond development subscriptions to production token usage, cloud hosting, and security risks. The report concludes that AI cost concerns have not substantially slowed adoption, but that stronger governance, security, and operational visibility will increasingly shape investment decisions.

Effective AI governance depends on answering three runtime questions: who is acting on data, what resources and actions they can access, and whether the organization can maintain control while agents operate. The framework distinguishes human users, agents borrowing human permissions, and service agents with independent identities, arguing that each requires deliberately scoped least-privilege access and clear attribution. It recommends defining granular policies centrally across environmental, platform, and application layers, then enforcing them consistently at runtime rather than rebuilding controls within each app. Control should rely on deterministic rules that models cannot override, comprehensive audit trails, potential guardian agents to detect policy violations, protections that keep sensitive data within organizational boundaries, and spending limits tied to use cases and outcomes. Citing a survey in which few technology leaders rated their governance or production visibility as strong, the piece urges organizations to begin with a high-value workflow, assess it against the three questions, and select platforms that provide unified governance across apps, automations, and agents.