Research library
ZAN-RN-015Concept noteResearch question

Local-first AI assurance sandbox

A concept note scoping a unified local sandbox environment that hosts the control plane's evaluation, trace, and security modules for fully offline assurance testing.

local-firstsandboxoffline evaluationprivate inference
Technical uncertainty

Whether a single local sandbox environment can host the evaluation harness, trace layer, and security harness together without conflicting resource or isolation requirements.

Expected evidence
  • Sandbox architecture sketch
  • Resource isolation model
  • Offline evidence export format
Next experiment

Stand up a single-machine sandbox running a local open-weight model, the evaluation harness, and the security harness together, and measure whether evidence output remains consistent with the cloud-connected control plane.

Hypothesis

A local-first assurance sandbox, built around a shared trace and evidence schema, can run evaluation, security, and cost-profiling modules against local open-weight models before any workflow is exposed to cloud infrastructure.

Scope

Why assurance testing should be able to run fully offline

Early-stage assurance testing on sensitive workflows often cannot depend on external model APIs. This note scopes what a local sandbox needs to host evaluation, tracing, and security testing together without cloud dependencies.

References

Related research