Free for students & researchers · No credit card required
Learn AI Governance
by doing it for real.
Gegentica is a hands-on learning environment for AI risk governance. Write real policies, detect real violations, and map controls to real compliance frameworks — in a safe sandbox built for education.
Learning modules
12live
Compliance frameworks
7mapped
Detectors to explore
64built-in
Time to first policy
<15min
01 / The Sandbox
A live governance console,
ready to experiment in.
The sandbox mirrors a real AI governance platform. Write a policy, fire a test prompt, watch violations surface in the dashboard — all without touching production systems.
gegentic / student-sandbox / violations
sandbox · learning mode
Violations
last 60 minutes · 9 violations · 2 blocked · 3 flagged for review
Live
History
Export
Test Requests
247
sandbox run
Violations found
9
High: 2
Policies active
5
2 drafts
Frameworks mapped
3
NIST · EU AI Act
Time
Event
Policy
Severity
Action
10:14:02
SSN in completion → test-agent · gpt-4o
PII / US-SSN
HIGH
BLOCK
10:13:48
Prompt injection · ignore-all-instructions
Injection
HIGH
BLOCK
10:12:31
Off-allow-list model · llama-3.1-405b
Model / AL
MED
REVIEW
10:11:55
Customer record in response → rag-agent
DLP / Customer
MED
REDACT
10:10:20
Toxicity 0.79 in agent reply
Content / Tox
LOW
LOGGED
02 / Capabilities
Six modules.
One complete picture of AI governance.
Each module teaches a distinct discipline — from writing your first policy to producing audit-ready evidence. Work through them in order, or jump to what you need most.
01 — WRITE
Policy Authoring
Author governance policies from scratch using a structured policy editor. Define criteria, set actions (block / flag / allow), and assign policies to AI agents in your sandbox.
Policy Sets · criteria & actions
02 — DETECT
Violation Lab
Fire test prompts against your agents and watch violations surface in real time. Explore 64 built-in detectors covering PII, prompt injection, toxicity, model allow-lists, and more.
64 detectors · live violation stream
03 — MAP
Framework Mapper
Link your controls to NIST AI RMF, EU AI Act, ISO 42001, SOC 2, and PDPA. Understand how a single policy decision satisfies multiple regulatory obligations simultaneously.
7 frameworks · continuous mapping
04 — REVIEW
Human-in-the-Loop Queue
Route borderline violations to a Review Queue. Practice the human-in-the-loop workflow — assign, annotate, and resolve — with a full audit trail from event to decision.
Review Queue · assign & resolve
05 — AUDIT
Audit Log Explorer
Every prompt, response, tool-call, and policy decision is recorded. Search, filter, and export the immutable log — exactly as a compliance auditor would expect to receive it.
Audit Log · searchable & exportable
06 — REGISTER
AI Use Case Registry
Maintain a living registry of the AI use cases and agents in your sandbox. Learn how governance attaches to workloads — not to calendars — by linking agents to their policy sets.
Use Cases · agents & prompts
03 / How it works
Governing your first
agent in under 15 minutes.
No prior experience required. The sandbox walks you through each step — and every action you take is real: real detectors, real violations, real audit trails.
STEP 01 · SIGN UP
Free account
Create your free account with a university or research email. Your personal sandbox is provisioned instantly — no billing, no approval queue.
STEP 02 · CONNECT
Hook up an AI agent
Route any OpenAI-compatible agent through the sandbox gateway. Works with any LLM provider — or use one of the pre-built demo agents to skip setup entirely.
Pre-built agents
3 ready to use
or bring your own OpenAI-compatible endpoint
→ agent connected
3 ready to use
or bring your own OpenAI-compatible endpoint
→ agent connected
STEP 03 · GOVERN
Write your first policy
Follow the guided policy editor to create your first Policy Set. Pick a detector, set an action, assign it to your agent, and fire a test prompt to see it work.
policy_set: no-pii
action: block
agent: demo-agent
framework: nist-ai-rmf
action: block
agent: demo-agent
framework: nist-ai-rmf
STEP 04 · EXPORT
Produce evidence
Export your audit log and framework mapping as a report. Use it for coursework, research papers, or to demonstrate governance fluency to a future employer.
audit log → 47 events
frameworks → NIST · EU AI Act
export → PDF · JSON
evidence ready
frameworks → NIST · EU AI Act
export → PDF · JSON
evidence ready
04 / Run it yourself
Prefer to run it
on your own infrastructure?
Gegentica ships as a self-contained Docker Compose stack. Spin it up on your own laptop, lab server, or university cluster — no external sandbox account required, and no data ever leaves your machine.
QUICKSTART · DOCKER COMPOSE
Clone & run
The gegentic-quickstart repo contains a ready-to-run docker-compose.yml that provisions the full sandbox stack locally.
# clone the quickstart repo
$ git clone https://github.com/gegentic/gegentic-quickstart
$ cd gegentic-quickstart
$ docker compose up -d
→ sandbox running at localhost:3000
$ git clone https://github.com/gegentic/gegentic-quickstart
$ cd gegentic-quickstart
$ docker compose up -d
→ sandbox running at localhost:3000
IMAGES · DOCKER HUB
Or wire up the images yourself
The sandbox is made up of several services. Each one is published as its own image under the gegentic Docker Hub organisation, versioned per release — browse the org to find the image and tag you need for your own Compose file or orchestrator.
05 / Who it's for
Built for everyone
learning responsible AI.
Whether you're studying AI policy, researching governance frameworks, or teaching the next generation of responsible AI practitioners — Gegentica gives you the tools to do it hands-on.
For Students
Go beyond theory. Govern a real AI agent.
- Write and test policies in a live sandbox
- See violations fire against real detectors
- Export audit logs for coursework and assignments
→ Computer science, AI ethics & law programmes
For Researchers
Experiment with governance models at scale.
- Test custom detectors and policy configurations
- Map novel frameworks to existing control taxonomies
- Export structured data for academic analysis
→ AI safety, policy & governance researchers
For Educators
Teach AI governance with a live platform.
- Provision class sandboxes with shared policy sets
- Design exercises around real violation scenarios
- Use the framework mapper as a teaching aid
→ University lecturers & bootcamp instructors
06 / The numbers
Everything you need to learn AI governance, at no cost.
Free
For students & researchers
Full sandbox access at no cost. No credit card, no time limit, no feature gating.
15m
Time to first policy
From sign-up to a running policy set with live violations — guided by the onboarding flow.
64
Built-in detectors
PII, prompt injection, toxicity, model allow-lists, topic drift, and more — all explorable in the sandbox.
7
Compliance frameworks
NIST AI RMF, EU AI Act, ISO 42001, SOC 2, HIPAA, PDPA, and internal AUP — pre-mapped and ready.
07 / About Gegentic
The company behind
the sandbox.
Founded in December 2025, Gegentic builds Gegentica — a middleware firewall that sits between AI agents and AI models, enforcing compliance, governance, and data sovereignty for healthcare, financial services, government, and enterprise deployments. This free sandbox is how we teach that same discipline.
Compliance you can trust
Automated enforcement of HIPAA, PCI-DSS, GDPR, PDPA, and government data protection laws.
Data sovereignty
Deploy on an internal edge server, cloud SaaS, or cloud PaaS — your data stays in your jurisdiction.
Governance & transparency
Every AI interaction is logged and auditable, so compliance teams can trust every decision.
08 / FAQ
Common
questions.
Something we didn't cover? Reach out and we'll get back to you within one business day.
Contact us →Is Gegentica really free?[01]
Yes. Gegentica is free for students, researchers, and educators. Create an account with your university or research email and your sandbox is provisioned immediately — no billing details required.
Do I need a background in AI or security to get started?[02]
No. The onboarding flow explains every concept as you encounter it. The sandbox is designed to teach AI governance from first principles — you learn by doing, not by reading a textbook first.
What AI models can I connect?[03]
Any OpenAI-compatible endpoint — OpenAI, Anthropic, Mistral, local models via Ollama, and more. You can also use one of the pre-built demo agents without connecting anything at all.
What compliance frameworks are covered?[04]
NIST AI RMF, EU AI Act, ISO 42001, SOC 2 (CC6/CC7), HIPAA, PDPA, and custom internal AUP. The framework mapper shows you which controls satisfy which clauses — and lets you add your own.
Can I use Gegentica for a university course or research paper?[05]
Absolutely. You can export the audit log, violation history, and framework mapping as structured JSON or PDF. Many students use these exports directly as coursework evidence.
How is the sandbox isolated from production systems?[06]
The sandbox is a fully isolated environment. No data you enter is shared with other users, and sandbox events never touch any live systems. You can experiment freely without any real-world consequences.
09 / Get started
The best way to understand
AI governance is to practise it.
Sign up for free, connect an agent, and have your first policy enforcing violations in under 15 minutes. No experience required — just curiosity.
Bridging the Gap between Innovation and Responsibility.
© 2026 Gegentic. All rights reserved.