Vercel AI Updates: August 21, 2026

1. Vercel Agent Comes to Slack Code Channels

Vercel. Vercel Agent can now operate within Slack code channels, allowing collaborative development workflows with approval gates. The integration brings agent actions into team chat, where developers can review and approve changes without leaving Slack. It positions the agent as a participant in existing team conversations rather than a separate tool. Source

2. Vercel for Slack Enters Public Beta

Vercel. Vercel announced Vercel for Slack, which entered public beta the same day. The integration brings Vercel workflows and deployment notifications into Slack, giving teams status updates and controls where they already communicate. It complements the Slack code channel support for Vercel Agent. Source

3. How v0 Authenticates to Snowflake Without Exposing OAuth Tokens

Vercel. Vercel published a technical post on how v0 connects to Snowflake without exposing a user’s OAuth token to generated code. The approach injects credentials at request boundaries rather than passing tokens into the generated application, reducing the blast radius if generated code is compromised. It is a useful pattern for anyone building AI code generators that touch authenticated data sources. Source

4. Fish Audio Models Added to Vercel AI Gateway for Free

Vercel. Vercel made Fish Audio models available on the Vercel AI Gateway at no cost. The addition expands the set of audio models accessible through the gateway’s unified API. It gives developers another speech and audio option without separate provider integration. Source

5. Bun 1.4 Runtime Available in Vercel Functions

Vercel. Vercel Functions now support the Bun 1.4 runtime, which features a Zig-to-Rust rewrite and improved Node.js compatibility. The update gives developers a faster JavaScript runtime option for serverless functions. Better Node compatibility lowers the friction of moving existing workloads onto Bun. Source

6. Custom Metrics Supported in Vercel Observability

Vercel. Vercel added support for custom metrics in Vercel Observability, letting developers emit custom metrics from Vercel Functions and analyze them alongside built-in observability data. This helps teams instrument application-specific signals, including token usage or model latency for AI features, without a separate monitoring stack. Source