How Featured Handles 100,000 Monthly Media Pitches on Vercel

AI copilot for the PR industry Featured has maximized operational efficiency through its Vercel migration, processing over 100,000 media pitches per month. A three-person engineering team supports more than 100,000 users and three brands, achieving the milestone of connecting users and journalists every six seconds.

Infrastructure Modernization and Operational Efficiency

The existing AWS Elastic Beanstalk environment imposed a heavy operational burden due to custom AI provider integrations and a separate orchestration system. The inefficiency of manual provisioning became particularly clear when 374 Sanity sites needed to be launched simultaneously, prompting the switch to Vercel. This enabled single-click deployments and centralized rollouts.

Leveraging AI SDK and Workflow SDK

Featured uses the AI SDK and AI Gateway to manage traffic across 17 models through a single abstraction. Custom rate-limiting and API integrations were removed, allowing model swaps to be handled via configuration changes alone without code rewrites. Additionally, long-running backend tasks (such as media monitoring and opportunity qualification) were migrated to the Workflow SDK instead of relying on separate systems.

Chat-Centric Interface and Development Acceleration

The company adopted the Vercel open-source agent framework eve to transition to a chat-centric interface. By leveraging the useEveAgent hook, which provides durable sessions, streaming, and tool calls out of the box, the rebuild was completed in one week—far faster than the initially estimated several months. The future goal is to expand the AI layer from PR into other sectors such as legal and finance.