Z7-Guard™ private AI for institutions
Private AI for the records you are responsible for.
Z7-Guard is private AI for institutions that cannot hand what they hold to a public model. Deploy on premises or inside your own tenant, so every conversation stays within a boundary you already control and already answer for.
- On-premises or your own tenant
- SLED
- Education
- Compass
Z7 Guard Compass: the Education application.
Z7 Guard is the private AI foundation. Compass applies it to K-12 and exceptional student education: student context and assessments in, reviewed plans, behavior reports and family documents out.
Z7 Guard for Business
Private AI for law firms and commercial teams, deployed inside your own boundary. Start with the client documents and workflows your team needs to protect, then scope the models, source citations, access controls and human review around that work.
Z7-Guard Legal is the first business use case. Infrastructure, integrations and supported workflows are defined for each deployment.
What to decide before you begin
A useful engagement starts with a specific outcome and clear responsibilities.
-
Choose the institutional workflow
SLED covers state and local government and higher education. Education covers K-12 and ESE. Start with a staff workflow, its record owners and its review requirements.
-
Make access demonstrable
Define which roles can retrieve which records, how retention works and who can administer the deployment. Test forbidden access before expanding the pilot.
Use the deployment checklist -
Keep decisions with people
Qualified staff remain responsible for student and institutional decisions. Agree on source checks, escalation, evaluation and the support model before rollout.
Private AI inside a boundary you already govern
Same principle in every lane: the AI comes to your data, your data never goes to the AI.
-
Your infrastructure, your boundary
Runs on premises or inside the Azure tenant you already govern. Nothing has to cross into a vendor's cloud for the assistant to be useful.
-
Built for sensitive records
A privacy layer sits in front of the records you are accountable for, rather than trusting a general-purpose model to behave.
-
US-origin models
For government and higher education deployments the underlying models are US-origin, so provenance is answerable when procurement asks.
-
Made to be reviewed
What the assistant produced, and why, can be shown to an administrator, a parent, or a reviewer.
The obligation does not transfer
When an institution puts its records into a public model, the accountability stays with the institution. Z7-Guard is built around that fact.
- 01
Scope your lane
Tell us which lane you are in and what you are accountable for, and we scope a deployment that keeps it there.
- 02
Deploy in your boundary
On premises or into your own tenant. The deployment lives where your governance already reaches, so there is no new boundary to defend.
- 03
Keep people in control
Human oversight is a design requirement, not a setting. The system supports the decision maker instead of replacing them.
Built for the institutions that carry the most obligation
Two lanes, each deployed inside the organization's own boundary.
-
SLED
State and local government and higher education
An AI assistant that keeps every conversation inside your own infrastructure, with a privacy layer built for the records an agency is accountable for.
- On premises or your own Azure tenant
- US-origin models
- No consumer AI service in the path
-
Education
K-12 and ESE
Private AI for K-12 school systems and the teams supporting ESE students, built so student records never leave infrastructure the district controls.
- District-controlled deployment
- Built around the ESE documentation load
- Human oversight required
- Engineered and operated by Z7
- On premises or your own tenant
- In business since 2012
- 24/7 SOC, NOC, and help desk
- SBA 8(a), MWBE, and DBE certified
- ISO 9001, 27001, 20000, 22301 aligned
- GSA MAS 47QTCA26D000F
Z7-Guard questions
Decisions, scope and operating responsibilities.
How is Z7-Guard priced?
Pricing follows the deployment scope, users and support needs. Book a call to define the requirements and receive a scoped proposal.
Can we start with a pilot?
Pilot scope, duration and acceptance criteria are agreed on a call before work begins.
Who controls our data and retention?
Your organization owns its instance and data. Models and harnesses are leased. Your content does not train shared models; access and retention requirements are agreed for your deployment.
Does Z7-Guard fit Google Workspace for Education?
We assess how Google Workspace for Education fits your identity, documents and workflow requirements. Integrations and permissions are confirmed in the deployment scope.
Does Z7-Guard make IEP or student decisions?
No. Qualified staff review AI-assisted work and retain responsibility for IEP and other student decisions. The system supports human judgment and oversight.
Does a private deployment automatically satisfy our regulatory obligations?
No. Privacy, records handling, permissions, contracts and oversight still need review for your institution. Deployment location is one control within a broader program.
What should we bring to the first conversation?
One staff workflow, its data categories, the intended users, your identity platform and the decisions a person must retain. Use approved examples without unnecessary sensitive records.
Bring AI inside your walls
Tell us which lane you are in and what you are accountable for, and we will scope a deployment that keeps it there.