Build voice checks and action rules into your AI support workflow. Sherlock connects caller evidence to policy decisions and execution controls. Your team can review why an action went ahead.
Check the caller’s voice.
Apply rules to the request.
Review the action taken.
THE BUILDING BLOCKS BEHIND A CHECKED SUPPORT ACTION.
Call supported checks from your existing support app, connect them in the visual editor, or work with our engineers on implementation.
Call voice verification and policy evaluation from your own code. Use a scoped API key for supported module operations while keeping orchestration in your app.
Connect the supported checks your agent needs. Configure their inputs and publish a versioned workflow with its own API URL and scoped key.
Work with an engineer to map your support workflow, define its action controls and build the integration around your requirements.
Map the path from a customer conversation to a protected action. Connect voice evidence to policy checks and inspect each step before publishing supported operations.
Add modules from the searchable picker and connect compatible inputs and outputs.
Pan, zoom, configure each node and validate the structure of your workflow.
Save locally in this browser or import and export your workflow as JSON. Publish supported operations as a system with its own API URL and key.
Save a draft before publishing supported operations.
Explore how your support workflow brings voice checks and action rules together. Preview the dashboard with clearly labeled sample activity.
Inspect the local graph inventory and explore sample system activity.
Voiceprint separates authenticity from identity. Policy shows decisions. Gate shows permits and replay controls.
Graph counts reflect your local design. Activity, consumption and allocations are explicitly labeled samples.
See how caller evidence, sign-in and action rules contribute to a support decision.
Voiceprint checks voice authenticity and speaker identity independently. Pulse brings together scores, trust budget and decision context.
Voice evidence informs a decision. It does not grant access.A support model can propose an action. Policy checks the request against explicit rules. In the demo, Gate controls execution and Ledger records the decision and outcome.
Deterministic rules can allow, step up, hold or request more speech based on the evidence and action.
A short-lived permit is tied to the exact action parameters and can only be consumed once.
The demo ledger links decisions, evidence references and outcomes in a verifiable hash chain. Raw audio stays outside the ledger.
Collect authenticity + identity evidence
Authenticate the principal
Authorize the exact action
Verify and consume the permit
Record the decision and outcome
Bridge executes sandbox transfers in the demo. Relay supports signed delegation between components owned by the same account. Remote connectors remain planned.
Explore the nine services behind conversation handling, voice checks and controlled actions. Each reference explains the available operations and current limits.
Read the module documentationA human or AI conversation, with trust checks in the background.
AI demo availableIndependent voice authenticity and speaker identity evidence.
Provider adapterTrust budget, decisions, evidence and stage latency.
Demo signalsAuthenticate principals and request action-bound step-up proofs.
Workspace sign-in + demo step-upExplicit rules decide whether an action can proceed.
Demo rules availableDelegate scoped tasks with receiving-side authentication and authorization.
Signed component delegationSingle-use execution permits bound to exact parameters.
Demo permits availableConnect authorized actions to your tools and APIs.
Bank demo + guarded sandboxAn evidence trail of decisions, handoffs and outcomes.
Demo audit availableStart in the visual editor. Connect the checks your support agent needs before it can act.