AI Governance
Built In, Not Bolted On
The governance-native AI platform for regulated industries. Every LLM call and every multi-agent pipeline is guided by APRA, ASIC, AFSL, AUSTRAC, OAIC, ISO 42001 and OWASP Agentic Top 10, audited in real time, and deployed inside your own VPC — your data stays where it's always been: with you. Helping you bring AI to production with confidence.
CPS230 APP.11 ASI02.6The Solution — Four Modules
Discover · Assess · Enforce · Audit — one bundled platform, four buyer outcomes.
Discover
Every AI surface, captured at the gateway.
- •Multi-provider chatbot SDK — OpenAI, Claude, Gemini, Fireworks, Ollama
- •Every LLM call and multi-agent turn captured before it reaches the model
- •Shadow-AI inventory across Copilot, Salesforce, internal LLMs
Assess
Risk-graded against 300+ controls in 60+ frameworks.
- •Policy-as-data catalogue · SCD2-versioned · auto-refreshed weekly by TinyFish
- •Multi-agent-native Gate → Main → Validator pipeline (AG2-compatible)
- •OWASP Agentic ASI07 / ASI08 / ASI10 enforced deterministically per turn
Enforce
Cited refusals, live — regulator names appear inline.
- •Blocks before the model when a request falls outside policy
- •“Based on APRA CPS 230 and OAIC APP.11…” shown in the chat response
- •Demonstrable in a 30-second pilot — not slideware
Audit
Auditor-ready evidence on whichever BI you already use.
- •Postgres operational store + S3 Iceberg immutable archive
- •dbt models for Tableau / Power BI / Looker / Metabase / in-house BI
- •Point-in-time reconstruction via SCD2 — “our APRA posture on 2026-02-15” is a SQL query
What's Special About Us
Eight things competitors can't say today.
Cited refusals, live
When a request falls outside policy, regulator names appear inline in the response.
- •e.g. “Based on APRA CPS 230 and OAIC APP.11…”
- •Something you can try in a 30-second demo
Policy-as-data
300+ controls across 60+ frameworks in a versioned catalogue.
- •Every decision joins to the control that fired
- •“What was our APRA posture on 2026-02-15?” is a SQL query, not a forensics project
Weekly regulator auto-refresh
TinyFish web-AI keeps the catalogue aligned with the regulator.
- •Scrapes APRA / ASIC / OAIC / OWASP every week
- •Diffs against catalogue, proposes reviewable changes
- •You ship against today's regulatory state, not last year's
Per-turn audit evidence
Write-once audit trail for every gate decision.
- •Postgres live store + S3 Iceberg immutable archive
- •Point-in-time reconstruction via SCD2
- •ISO 27001 and ISO 42001 aligned out of the box
Multi-agent-native
Governance pipeline IS itself a multi-agent system.
- •AG2-compatible Gate → Main → Validator pattern
- •Governs solo LLM calls AND multi-agent workflows
- •OWASP ASI07 / ASI08 / ASI10 enforced deterministically
Scheduler & warehouse agnostic
Fits the stack you already have.
- •Works with BMC Control-M and your existing scheduler
- •Works with your BI tool — Tableau, Power BI, Looker, Metabase, in-house
- •Reads and writes Postgres, Snowflake, BigQuery, Redshift
Bundled, not bolted
Governance is included with the platform, not a separate SKU.
- •One docker-compose deployment
- •One contract, one invoice, one support line
- •No extra GRC module to license or integrate
ISO 42001 in-house
Customer DDs are handled by our in-house compliance team.
- •Chief Compliance Officer holds ISO/IEC 42001 qualification
- •Gap assessments and audit prep done internally
- •Evidence packs tailored for each customer's framework
Deploys In Your Own Infrastructure
We ship a container that runs in your VPC, reads from your warehouse, writes to your S3 — so your data stays where it is today. Designed to align with APRA CPS 234, OAIC APP 8, and AU data residency from day one.
Your credentials, your keys, your data — always yours. We work inside the perimeter you already trust.
Your VPC
Runs as a container inside your existing VPC, subnets, security groups and WAF — no new network perimeter to approve.
Your Cloud Account
AWS, GCP, Azure, or on-premise — your account, your billing, your IAM. Credentials stay with you.
Your S3 / Object Store
Audit evidence (Iceberg archive) writes into your S3 bucket under your KMS keys. Access stays with your team.
Your Data Warehouse
dbt targets your Snowflake / BigQuery / Redshift / Postgres directly — data flows within your warehouse, not through us.
Your Orchestration
Works natively with BMC Control-M and your existing scheduler — our jobs run as standard shell commands, no bespoke runner required.
Your IdP & KMS
SSO via Entra, Okta, or Cognito. Encryption with your CMK — keys stay under your control throughout.
Data Stays Onshore
Deployed into AU regions with no cross-border transfer — aligned with APRA CPS 234 and OAIC APP 8 by architecture.
Air-Gapped Ready
Offline install packages available for classified environments. No outbound telemetry, no phone-home.
Frameworks Covered
60+ regulatory & industry frameworks · 300+ controls · auto-refreshed weekly
🇦🇺 AU Regulators
- APRA CPS 230
- APRA CPS 220 — Risk Management
- APRA CPS 234 — Information Security
- APRA CPG 235 — Managing Data Risk
- ASIC REP 798 + s912A
- ASIC AFSL — s911A / s961B Best Interests Duty
- AUSTRAC AML/CTF Act + AI/ML guidance
- OAIC Privacy APPs
- AU 6 Principles
- NSW AIAF
- TGA SaMD
🌏 International
- NIST AI RMF 1.0
- NIST AI GenAI Profile (NIST AI 600-1)
- NIST CSF 2.0
- ISO/IEC 42001:2023
- ISO/IEC 27001:2022
- UK 5 Principles (DSIT)
- FCA DP5/22
- EU AI Act
🇺🇸 US Sector
- FINRA Notice 24-09
- Colorado SB24-205
- NY DFS AI Guidance
🛡 Industry
- OWASP Agentic Top 10
- OWASP LLM Top 10
- NIST SP 800-53
APAC-native AI governance
Credo, OneTrust and Holistic AI are built US-first (NYC LL144, Colorado SB21-169) and EU-first (EU AI Act). We're built for this side of the Pacific — regulators your CCO actually answers to.
TinyFish auto-refresh treats new regulators as new scrape targets, not new code — APAC coverage extends as customers land.
Who It's For
Regulated verticals where AI governance is procurement-gating.
Financial Services
Banks · insurers · super funds · asset managers · advisers
- •APRA CPS 230 — operational risk
- •ASIC REP 798 — governance gap
- •AFSL obligations — s912A efficient, honest, fair
- •Prudential AI risk — accountable person
Healthcare
Clinics · hospitals · digital health · pharma
- •TGA SaMD — software as a medical device
- •OAIC APPs — health records privacy
- •AU 6 AI Principles — clinical decision support
- •State health dept AI guidelines
Government
Federal · state · local · agencies
- •NSW AIAF — AI Assurance Framework
- •QLD AI Governance · VIC AI Principles
- •AU 6 Principles — contestability
- •Public accountability obligations
Ready to see it live?
90-second demo. Live chatbot citing APRA, ASIC, OAIC in your own test session. No slideware.