Vercel says AI agents handle 91% of support tickets and inbound sales

Vercel has publicly stated that it has automated its inbound sales pipeline and that artificial‑intelligence (AI) agents now resolve the majority of its support tickets. The company’s central European head, Julian Eckerle, explained that the internal adoption of these agents began roughly two and a half years ago, marking the start of a systematic rollout within the organization.

The rollout followed a staged approach. Initially, Vercel encouraged its employees to experiment with AI tools and to construct their own agents. After this exploratory phase, the firm introduced a governance layer that includes identity management, logging, security rules, and token‑cost monitoring, thereby formalising the operational environment for the agents.

From Manual Teams to an Autonomous Sales Agent

According to Eckerle, a single autonomous sales agent now works together with one human to handle an inbound sales flow that previously required the effort of more than ten staff members. This reduction in headcount for the same workflow is presented as a core efficiency gain for the company.

Eckerle provided a cost figure for operating this autonomous sales pipeline, estimating annual expenses at about 5,000 USD. He also noted that the volume of the pipeline has more than doubled over a twelve‑month period. Both the cost estimate and the growth claim have not been independently audited, which limits the ability to verify the magnitude of the reported savings.

Support Ticket Automation and Workforce Redeployment

Vercel claims that its AI agents automatically resolve 91 % of support tickets. The company further states that employees who previously handled these tickets have been reassigned to more complex tasks rather than being laid off, suggesting a shift in job responsibilities rather than a net loss of staff.

The firm also acknowledges that its overall hiring may have decreased as a consequence of the automation. This admission highlights a potential indirect impact on recruitment patterns, although no specific hiring figures are disclosed.

Vercel warns that a blanket directive to use AI without careful case‑by‑case selection could generate frustration among staff. This caveat points to the importance of aligning AI deployment with suitable use cases and managing employee expectations.

Technical Controls via Vercel AI Gateway

The Vercel AI Gateway documentation confirms that the platform includes mechanisms for budget enforcement, access policies, cost tracking, latency monitoring, token usage limits, and routing attempts. These controls are intended to provide operational transparency and to mitigate risks associated with uncontrolled AI usage.

By integrating such controls, Vercel aims to balance the flexibility of AI agents with safeguards that prevent cost overruns and security breaches. The presence of these features, however, does not automatically guarantee that all deployments will adhere to best‑practice governance.

• Automation of inbound sales flow
• Resolution of 91 % of support tickets
• Implementation of identity and cost governance
• Reallocation of staff to higher‑complexity tasks

The case presented by Vercel illustrates a declared capacity gain rather than a universally applicable proof that AI agents can replace commercial or support services without risk. The company's statements are framed as internal observations, and they do not constitute evidence that similar outcomes would be achieved in different organisational contexts.

Open questions remain regarding the reproducibility of the reported cost savings, the long‑term impact on employee skill development, and the scalability of the governance model as the volume of AI interactions grows. Further independent evaluation would be required to assess whether the observed efficiencies persist under varied operational conditions.

In summary, Vercel’s experience suggests that, under a structured rollout and with appropriate governance, AI agents can markedly reduce the human resources needed for specific commercial and support processes. Nonetheless, the evidence is limited to internal metrics, and broader claims about the universal substitutability of human agents remain speculative.