Workflow SDK now compresses run and step payloads
85% storage savings. Vercel's Workflow SDK now auto-compresses AI agent payloads—no code changes needed.
Why it matters
Vercel ships infrastructure optimization that directly reduces operational costs for AI agents built on its platform. For founders running high-volume agentic workflows, this translates to measurable cost savings and faster execution without engineering lift.
The key facts
13 to knowWorkflow SDK @5.0.0-beta.19 introduces automatic zstd compression
Storage size reduction up to 85% for typical AI conversation JSON payloads
Example: 52 MB uncompressed workflow reduced to 10 MB with compression
Compression applies to run/step inputs, outputs, and conversation history
Compression is adaptive—only applies when beneficial, small payloads pass through uncompressed
No code changes required; applies automatically to eve agents and existing workflows
Workflow SDK beta 5 adds automatic zstd compression for run, hook, and step payloads
Compression is intelligent—only applies when beneficial, leaves small payloads unchanged
Storage size reduction up to 85% for typical JSON payloads from AI conversations
Example: 52 MB workflow compressed to 10 MB
Applies to eve agent conversation history and state persistence
Requires update to @5.0.0-beta.19 or later
No code changes required for existing workflows
Go to the source
Vercel Blogvercel.com
Publisher excerpt: The 5 beta now compresses all run, hook, and step inputs and outputs with .Workflow SDKzstd Compression kicks in automatically, but only when it helps. Small payloads stay as-is, larger ones get compressed before they're persisted. Compressed payloads use less storage and are faster to read and…