Casework triage
- Routes and prioritizes constituent cases against typed Case objects, with restricted personal data locked at the property level.
- Ontology permissions
AI agents for the public sector
Casework, eligibility, and constituent service — with full transparency.
Government agents handle casework, eligibility determinations, and constituent service with auditable decisions, records retention, and an always-armed kill switch. When a citizen, an oversight body, or a FOIA request asks how a decision was made, the answer is a complete, verifiable record.
Aligned with NIST AI RMF · FedRAMP-style controls · FOIA · records retention
Ungoverned sprawl
no identity · no audit · no caps
Governed runtime
identity → gates → ledger
The regulatory pressure
A public decision must be explainable to the person it affects.
Government AI carries a duty of transparency, due process, and records retention that private systems don't: an eligibility denial must be explainable, a record must survive a FOIA request, and an oversight body must be able to inspect the process. An opaque agent is a due-process problem. Cortex makes every determination auditable, retains the record, and keeps a kill switch armed for the whole fleet.
NIST AI RMFFedRAMP-style controlsFOIA · transparencyRecords retentionDue processISO 27001
Regulatory pressure
NIST AI RMFFedRAMP-style controlsFOIA · transparencyRecords retentionDue processISO 27001
Typed objects
Restrictedlocked
frameworks ▸ encoded as policy · gated · sealed
Use cases
Public-facing work, with the record built in.
Each agent maps the work to typed objects, runs under transparent oversight, and seals every determination into a retainable, FOIA-ready record.
Eligibility determination
- Assembles eligibility findings grounded to the application and the rules, with the determination pending human review — never an unexplained auto-denial.
- Oversight modes
Constituent service
- Answers inquiries grounded to public records and policy, with disclosure language enforced by policy so responses stay accurate and consistent.
- Policy-as-Code
Records & FOIA assist
- Assembles records for retention and FOIA response, where every assertion cites its source and the full decision lineage is available on request.
- Trust Ledger
Fleet-wide kill switch
- An always-armed kill switch lets an operator pause every agent instantly — essential when a public service must stop on a moment's notice.
- Control Tower
Transparency and due process by construction
Every determination explainable, retained, and inspectable.
A government decision can't be a black box. Cortex grounds each determination to its inputs, holds eligibility decisions for human review, retains the record, and exposes the full 10-hop lineage — so a constituent, an auditor, or a FOIA officer gets a complete, verifiable answer.
- Eligibility determinations pend for human review — no unexplained auto-denial
- Every decision retained and sealed for records-retention and FOIA
- Full 10-hop lineage available for oversight-body inspection
- Always-armed kill switch pauses the whole fleet instantly
Ungoverned sprawl
no identity · no audit · no caps
Governed runtime
identity → gates → ledger
Prove it — don't just claim it
The verdicts an oversight body will see.
These are the literal runtime responses — an opaque or unsourced determination fails closed.
runtime · verdicts
Eligibility denialHOLD · review
Unsourced finding451 blocked
PII read403 redacted
FOIA record200 sealed
fail-closed ▸ challenge it and it denies, on the record
- 01
Model the domain
Map the work to typed ontology objects — Constituent, Case, Application, Determination, Record — with restricted fields like ssn / benefit_detail locked at the property level.
- 02
Encode the rules
Translate the obligations above into Policy-as-Code and Action Fabric approvals — the gates fail closed, returning a precise code, not a guess.
- 03
Prove the outcome
Every run lands in a hash-chained Trust Ledger with signed receipts and 10-hop provenance — a record you can hand the regulator.
10-hopLineage on every determination
1 clickKill switch to pause the fleet
100Live-verified run quality score
256Chained SHA-256 sealing every record
Works with the rest of the runtime
The controls behind public sector.
The same shared runtime powers every vertical — here are the capabilities and the Solution Pack this industry composes from.
Control Tower
Watch every agent live and pause the whole fleet with one armed kill switch.
Oversight modes
Hold eligibility determinations for human review by default.
Trust Ledger
Retain and seal every determination for records retention and FOIA.
Ontology
Model Case, Application, and Determination; lock personal data fields.
Policy-as-Code
Encode eligibility rules and disclosure language as enforced policy.
Trust Center
Map auditability and retention to NIST AI RMF and FedRAMP-style controls.
Security & compliance
Built for the transparency and retention government owes the public.
Auditability, retention, RBAC, and multi-tenant isolation are shared across casework, eligibility, and service — so you map once, to NIST AI RMF and FedRAMP-style controls. Aligned with, never claiming an authorization you don't hold.
NIST AI RMF · GovernFedRAMP-style controlsFOIA · transparencyRecords retentionDue processISO 27001 · integrity
FAQ
AI agents for public sector — questions, answered.
How does Cortex make a government AI decision explainable?
Every determination is grounded to its inputs and sealed with a 10-hop provenance graph from the human who initiated it to the outcome. A constituent, auditor, or FOIA officer can be shown the complete, verifiable basis for the decision.
Can an AI agent deny benefits or eligibility on its own?
No. Eligibility determinations run under oversight modes that hold them for human review — there is no unexplained auto-denial, and the reviewer's decision is sealed into the record.
How does Cortex support records retention and FOIA?
Every decision and its lineage are retained and sealed into a tamper-evident ledger, so records survive a FOIA request with their full, verifiable provenance intact.
Is there a way to stop all agents immediately?
Yes. An always-armed kill switch in the Control Tower lets an operator pause the entire agent fleet instantly — important when a public service must halt on a moment's notice.
More industries
One governed runtime, every regulated vertical.
The controls are the same — the obligations they satisfy differ. Explore another vertical, or see the full set.
Answer 'how was this decided?' with a complete record.
See auditable determinations, records retention, and the always-armed kill switch behind every public-sector workflow.