Cursor claims half the Fortune 500, Agentforce 18,000 companies; neither discloses an audit trail or AI liability terms

Cursor, a coding agent, says on its site: "Cursor runs in your terminal, collaborates in Slack, and reviews PRs in GitHub." The company adds it is "Trusted by over half of the Fortune 500 to accelerate development, securely and at scale." Salesforce's Agentforce says it "can reason through complex requests, safely retrieve accurate knowledge about your business, take action on your behalf, and scale across teams and departments," and reports: "Over 18K companies already run on Agentforce." Both entities cleared a field-by-field audit this week and were promoted to aidispatch.news's published index.

On oversight, each publishes exactly one mechanism. Cursor's site says its "Agents use their own computers to build, test, and demo features end to end for you to review" — a check that happens after the work is finished, the only such point the page describes. Salesforce says Agentforce's agents "escalate the matter to human agents" only "When faced with complex issues beyond their scope" — a fallback for failure, not a standing limit on what an agent is allowed to do. Neither site, as checked, describes a mechanism that enforces scope, spending or authority limits while an agent is working.

Neither company publishes a retained, reviewable record of what its agents did, and neither names an independent safety evaluation. Salesforce points instead to an analyst ranking, saying "Gartner named Salesforce a Leader in the 2026 Magic Quadrant for Conversational AI Platforms" — a placement on an analyst quadrant, not an audit or certification of the system itself. Cursor's site names no certification, regulatory alignment or third-party evaluation at all.

The commercial terms that would apply if an agent's work fails are also undisclosed. Cursor's site publishes no pricing information. Salesforce says "Every Salesforce customer can get started with Agentforce for free, for any use case, with Salesforce Foundations," and otherwise prices through "Flex Credits, Conversations, or per-user licensing" — structures that charge a set amount regardless of whether an agent's output is used. Neither company publishes an AI liability insurance policy, a named carrier, a customer indemnification term or a liability cap.

The two products differ in almost every other respect — one an individual developer tool, the other an enterprise platform Salesforce sells at the department level — but converge on the same gap. Across deployments the companies count in the hundreds of thousands of engineers and more than 18,000 businesses, the trust-gap dimensions aidispatch.news tracks — audit trail, compliance, insurance and indemnity — go undisclosed on both.