Vercel

Vercel AI SDK Audit Logging & Evidence

Edge and web AI with evidence at the middleware layer

The Vercel AI SDK powers most TypeScript web and edge AI deployments. Sigigo middleware captures stream boundaries, tool calls, and model metadata without leaving your route handlers.

Where Sigigo sits

At the Vercel AI SDK middleware boundary — wrapping generateText, streamText, and tool-enabled flows before responses reach the client. Model provider details are captured for GPAI traceability.

Web and edge deployments

Most EU-facing copilots and customer-facing AI features ship on Next.js and edge runtimes. When high-risk Article 12 logging applies, automatic recording should attach at the same boundary where inference happens — not only in batch backends.

  • Stream start/end timestamps for decision windows
  • Tool call metadata tied to the user session
  • Commitment mode keeps prompts customer-held on edge workers

Integration shape

@sigigo/integrations-vercel-ai — middleware wrapper composable with existing provider clients and LangChain stacks.

Evidence events

Canonical event types emitted through this integration path:

  • ai_act.inference_started / inference_completed
  • ai_act.input_received (commitment)
  • ai_act.output_generated (stream completion)
  • ai_act.action_executed (tool invocations)

Related

Integration availability may vary by tenant, region, and deployment model. This page describes product architecture — not a guarantee of feature availability. See compliance disclaimers and contact us for early access.