Getting started
Set up your first assurance run, invite your team, and produce your first report.
Help center
How can we help?
Guides, answers, and contacts for everything from your first assurance run to enterprise rollout. If you can't find what you need, our team is one email away.
Getting started
Connect your first model, run a test, and produce your first framework-aligned report - step by step.
Connectors & integrations
GitHub, Jira, Slack, Deepgram, Webhooks, and SSO. Wire AgenticAssure into the stack you already run.
External Auditor Seats
Time-boxed, read-only access for Big-4, Notified Bodies, MAS, and NCAs - with cryptographic seal and RFC3161 timestamps.
Browse by topic
Find the answer, fast.
Eight categories cover everything from running your first test to granting External Auditor Seats.
Platform & testing
Run regulatory-grade red-team tests across LLMs, agents, and tool-using systems.
External Auditor Seats
Grant time-boxed, read-only access for Big-4, Notified Bodies, MAS, and NCAs.
Compliance & frameworks
Map your evidence to 12 frameworks - OWASP, NIST, MITRE, MAS, AI Verify, EU AI Act, HIPAA, CCPA, and more.
Reports & evidence
Generate, export, and share framework-aligned evidence packs your auditors will accept.
Account & billing
Credits, invoices, seat management, and how to upgrade or change your plan.
Integrations
Connect AgenticAssure to your CI/CD, ticketing, SIEM, and identity provider.
Privacy & security
How we handle your prompts, model traffic, and customer data - in transit and at rest.
Step-by-step
Detailed guides
What is AgenticAssure?
AgenticAssure is the governance and assurance platform for enterprise AI. Six modules - Discover, Control, Test & Prove, Govern, Analysis, Assurance - cover your AI estate from inventory through runtime enforcement to continuous compliance monitoring.
Most teams use AgenticAssure to get from "the model works" to "the model is safe to ship and we can prove it to any regulator" in under a week.
Who is it for?
CISOs, AI governance leads, compliance directors, ML platform teams, and Big-4 audit partners at regulated enterprises - typically in financial services, healthcare, public sector, and critical infrastructure. Singapore MAS-regulated and EU AI Act-scoped organisations are a core vertical.
Run your first test
- 01Connect your model. Paste an HTTPS endpoint or pick a managed provider (OpenAI, Anthropic, Google Gemini, Ollama).
- 02Pick a probe set. Start with the OWASP LLM Top 10 baseline. Add NIST AI RMF, MITRE ATLAS, or MAS MindForge packs as needed.
- 03Run. The first run typically completes in minutes. Results appear in real time in the platform dashboard.
- 04Export the report. Framework-aligned evidence pack generated automatically. Export to PDF or JSON.
External Auditor Seats
Provision time-boxed, read-only access for outside auditors - Big-4, Notified Bodies, MAS examiners, NCAs. Access is scoped to verification endpoints only. Every seat carries a cryptographic seal, judge provenance, RFC3161 timestamp, and probe manifest.
Every auditor request is appended to the immutable audit log. Audit activity never touches your model's production traffic.
Continuous vs point-in-time
Point-in-time gives you a snapshot - useful before a launch or audit. Continuous Monitors re-run the relevant probes whenever your model, prompt, or tool surface changes, so your evidence never goes stale.
We recommend continuous for any system in production. Most enterprise customers run continuous on production and point-in-time on staging release candidates.
Framework-aligned reports
Each report covers: executive summary, scope, methodology, framework-by-framework findings, detailed probe results, severity-weighted scoring, recommended remediations, and a verification metadata appendix. Blockchain-anchored test results mean each section is independently verifiable.
EU AI Act runs include an auto-generated Annex IV Dossier. AIVTF runs produce official PDF + Annex A CSV in one click.
FAQ
Frequently asked questions
What does AgenticAssure actually do?
AgenticAssure discovers every AI system in your estate, enforces policy at runtime, tests against 34 attack techniques, and maps results to 12 regulatory frameworks - so your board, your CISO, and your auditors get evidence they actually accept. Six modules: Discover, Control, Test & Prove, Govern, Analysis, Assurance.
How long does an assurance run take?
A typical first run completes in minutes for an LLM endpoint and within hours for a full agentic system with tools. Continuous Monitors re-run incrementally on every model or prompt change, so you have fresh evidence whenever you need it.
Do you store our prompts or customer data?
Test traffic is processed in-region using your chosen data residency. Production traffic is never stored. Test artefacts (probes, responses, scores) are retained per your plan's retention policy (90d default) and can be exported or purged at any time. See our Privacy Policy for details.
Which models and providers do you support?
OpenAI, Anthropic, Google Gemini, Ollama (local), and any HTTP-accessible endpoint - 82 connected models in the live platform. Multi-provider fallback chain with mTLS + custom headers.
What are External Auditor Seats?
Time-boxed, read-only access for outside auditors - Big-4, Notified Bodies, MAS examiners, NCAs. Scoped to verification endpoints only. Every request carries a cryptographic seal, judge provenance, RFC3161 timestamp, and probe manifest. Every auditor request is appended to the immutable audit log.
Can I run AgenticAssure inside my own VPC?
Yes. Enterprise plans support fully self-hosted deployment in your VPC or on-premises, with the same managed update cadence as the cloud product.
How is AgenticAssure priced?
Credit-based. 1 credit = $50. Per-test costs range from 2 credits (Blockchain Logging) to 25 credits (Security Testing). Top-up tiers from $499 (Starter) to $39,999 (Enterprise). See the Pricing page.
How do I cancel or change my plan?
Use the Manage subscription link in the footer to update payment details, change plan, or cancel. Enterprise contracts are managed by your account team.
Where can I report a security vulnerability?
We run a coordinated disclosure programme. See our Security page for scope, reporting steps, and our commitments to researchers.
I'm stuck. How do I reach a human?
Email contact@agenticassure.ai or use the contact form. Enterprise customers also have a named technical account manager and a dedicated Slack channel.
Still need help?
Talk to a human.
We're a small senior team - your first reply will come from someone who has built and shipped this work for regulated enterprises. Most messages get a response the same business day.