Citizen data stays in the institution, every decision recorded.
Public institutions use language models for petition classification, legislation search, call centres and e-Government assistants. Presidential Circular 2019/12 and the Information and Communication Security Guide keep critical data in the country, on a closed network. Civil-service neutrality rules topics such as politics out of scope. Thulpar’s public-sector template ships the topic rail, citizen PII detection and the five-year retention floor ready to use.
tr-public-sector
Air-gap bundle or on-premise, Standard tier. The tr-public-sector template. The public-tr evidence profile.
- Presidential Circular 2019/12
- Information and Communication Security Guide
- KVKK 6698 · Law 5651
- Public classification scheme, 5-year retention floor
The closed-network mandate
Systems that process critical data cannot reach the internet. Cloud guardrails are not an option for them.
Watch the runs →The scenarios below are derived from the product’s real rules.
Every scenario is an agent run written to the product’s real rules. It shows the stage, the rails, the decision, the finding, the approval and the ledger entry step by step.
3 scenarios →What comes up in practice
The closed-network mandate
Systems that process critical data cannot reach the internet. Cloud guardrails are not an option for them.
Topic boundaries
A public assistant must not comment on politics or give investment advice.
Retention and audit
The Court of Accounts and internal audit require records kept beyond five years. Those records have to be immutable.
3 scenarios
The player opens on the flagship scenario. Pick the others below.
A citizen asks about an application; the TCKN is masked, a political question rewritten, the AI notice put first; all on the closed network.
9 GR-*
| id | control | OWASP | rails |
|---|---|---|---|
| GR-TRANS-01 | AI disclosure: chat + voice | — / ASI09 | conversation |
| GR-PII-01 | PII and special-category detection + masking (TCKN, IBAN, phone, health) on input/tool/output/log | LLM02 / — / MCP10 | pii.detect, pii.redact, file.metadata_pii |
| GR-CONT-01 | Content-harm policy (AILuminate 12 + TR) with tenant thresholds | LLM09 / ASI09 | content.input, content.output, toxicity, topic, groundedness, language, image.content_safety, banlist, competitors |
| GR-TOOL-02 | Durable human approval on high-impact actions, four eyes | LLM06 / ASI02, ASI09 | labels.sink |
| GR-LOG-01 | Tamper-evident event log (model/tool/policy/approval/hit), ≥ 6 months, auditor export | — / — / MCP08 | |
| GR-OUT-01 | Output handling: no raw HTML, code sandbox, schema validation | LLM05 / ASI05 / MCP05 | code, schema, regex, secrets, file.analyze, file.type, file.active_content, yara |
| GR-EXF-01 | Tenant egress allowlist; no remote images in chat; no PII in URLs; SSRF | LLM02, LLM05 / ASI02 / MCP06 | exfil, tool.args, file.qr |
| GR-MEM-01 | Memory-poisoning policy + provenance | — / ASI06 / MCP10 | memory.write |
| GR-IFC-01 | Information-flow labels (integrity, confidentiality, tenant, PII class) enforced at sensitive sinks | LLM02 / ASI01, ASI03 / MCP10 | labels.sink, pii.detect |
See it in your own environment.
A one-hour technical demo: your scenario, your policy pack, on a closed-circuit install. Then the install kit and a trial licence for a pilot.