Intercept the request
Route outbound commands through a governance gateway between the agent and the estate’s hardware.
SASKI Estate / Vertical application
The SASKI model.
Applied to the home.
De-risk AI-enabled smart homes with integrator-defined rulebooks and attestation for governance decisions.
For smart-home integrators
One governance architecture. A connected-home application.
Dimming a light, unlocking a gate, and changing a network port should not carry the same permissions.
SASKI Estate is a vertical application of the same independent governance used across SASKI products. It separates the AI’s reasoning from the authority to execute a physical action.
The integrator translates operational requirements into a rulebook defining which tools, targets, and parameters are allowed, which actions require a hold, and what evidence the attestation retains.
See SASKI Estate in action
A governed smart-home demonstration.
This demonstration is hosted by YouTube. Watch directly on YouTube ↗
The governed path
From a request to an accountable action.
01Homeowner request
A person describes the desired outcome or a problem to diagnose.
02AI proposal
The agent reads relevant context and proposes a specific action.
03SASKI rulebook check
Scope, parameters, prerequisites, and hold requirements are evaluated.
04Control-system command
Only a permitted action proceeds through the governed path.
05Device action & attestation
The equipment receives the command and the governance decision is retained for verification.
The governance architecture
Work with the installation you have.
Existing control systems
Add governed AI control and troubleshooting above existing automation systems. SASKI Estate works across control transports, so the governance boundary is independent of how a device receives its commands.
New installations
For a new build, SASKI Estate can serve as the control layer directly without a separate AV processor. Define the equipment, required integrations, and control scope for the installation.
Supported control systems
- Crestron
- Control4
- RTI
- Q-SYS
Before equipment acts
Apply the rulebook before execution.
Evaluate the rules
Check each tool invocation against deterministic governance, parameter bounds, and security constraints.
Control what proceeds
Block or modify unauthorized commands before equipment receives an execution signal. Apply required holds before proceeding.
AI-driven troubleshooting
Diagnose the real installation.
As-built system knowledge gives AI the context to investigate the home it is actually controlling.
Lab example
The right camera. The wrong input.
A camera mapped to HDMI 1 was displayed on HDMI 2. A troubleshooting request identified the mismatch and corrected the input through the governed path.
- The homeowner reports a symptom.
- AI reads the relevant live state.
- A trusted site model supplies installation-specific facts.
- AI identifies a likely mismatch or failure point.
- A corrective action is proposed.
- SASKI checks scope, bounds, prerequisites, and hold requirements.
- Only the approved action reaches the device.
Attestation
Explicit decisions. Verifiable records.
Fail-closed command validation
Unrecognized tools, undeclared parameters, or schema mismatches produce a deny decision. Coverage depends on routing the relevant commands through the governed path.
Cryptographic attestation
Ed25519-signed execution receipts record action context. External Firestore and KMS anchoring supports independent verification of the audit trail.
A foundation for ongoing service
Keep the integrator in control.
One governed backend for buttons, AI, automation, and technician tools.
Use installation knowledge as an operational asset and a more data-driven interface to reduce repetitive touch-panel programming.
Build recurring services around remote diagnostics, an AI concierge, governance policy, monitoring, and premium support.
Discuss SASKI Estate for your installations ↗Start with your use case