Home About Us 📋 About ARIA 🛡️ Launch Platform 🔍 AI Security Assessment 🔄 Security by Design 🛡️ AI Security 🤖 AI Safety & Guardrails 🌐 IoT Cybersecurity 🔒 Network Security 🤝 Engage with Aggi 🏥 Healthcare Careers Contact Us →
AI Endpoint Onboarding Pilot

Connect your AI endpoint.
See what ARIA surfaces.
15 minutes.

No contract, no commitment, no sales pitch. Register your LLM, run your first behavioral evaluation, set a drift threshold. If it's useful, we talk next steps. If it's not, you've spent 15 minutes and learned exactly where your model stands.

Start the Pilot Conversation → See the 5 Steps ↓
15minutes to first eval
6LLM providers supported
160behavioral fixtures
12active frameworks
$0contract to start
How It Works

Five steps. Under 40 minutes total.

Walk through these steps and you'll have a live evaluation baseline, a scheduled re-run cadence, and a drift alert threshold — all before your next meeting. Step 5 is optional; activate it when you want runtime enforcement on top of periodic testing.

1
10 minutes
Complete the onboarding wizard
Answer five questions about your industry, deployment, geography, data type, and audit posture. ARIA's rule engine — not an LLM — deterministically recommends the right framework set. You can override before committing.
2
5 minutes
Register your LLM endpoint
Name your endpoint, select your provider, enter your API key. ARIA encrypts it immediately on save — never readable again, including by Aggi staff. Six providers supported; self-hosted endpoints require HTTPS.
3
5 minutes
Run your first evaluation
Click Run eval. ARIA fires behavioral test fixtures against your LLM — does it refuse to echo PHI? Does it hedge clinical claims? Does it hold up against prompt injection? Results land in 30 seconds.
4
5 minutes
Schedule re-runs + set a drift threshold
Set a cadence (weekly is the standard starting point), pick an off-peak hour, and set a drift alert threshold. 10% is a reasonable starting point — it catches meaningful behavioral shift without over-alerting.
5
15 minutes
Activate the inline proxy
Optional. Instead of periodic testing only, every production LLM call runs through 7 pattern gates + 3 AI judges in real time. You change one URL and one API key in your application. Less than 15ms overhead.
Supported for OpenAI-compatible endpoints and self-hosted. Additional providers on the roadmap.
LLM Providers

Six providers. No vendor lock-in.

ARIA orchestrates against the LLM cloud your security team already approved. Your API key stays in your account — ARIA charges for orchestration, never per inference token.

Provider What to enter Shape C inline proxy
Anthropic https://api.anthropic.com Evaluation (Shape B) — proxy roadmap
OpenAI https://api.openai.com ✓ Live
Amazon Bedrock Your AWS region, e.g. us-east-1 Evaluation (Shape B) — proxy roadmap
Google Vertex AI Your GCP region, e.g. us-central1 Evaluation (Shape B) — proxy roadmap
Azure OpenAI https://<your-resource>.openai.azure.com Evaluation (Shape B) — proxy roadmap
Self-hosted Base HTTPS URL of your endpoint, e.g. https://vllm.example.com/v1 ✓ Live
Framework Selection

Which frameworks will ARIA test against?

The onboarding wizard recommends the right set for your situation. Here's what that typically looks like by industry. You can always override the recommendation and select your own combination.

Industry / situation Typical framework set ARIA recommends
US healthcare / clinical AI NIST AI RMF · HIPAA · FDA CDS · ISO 42001
EU medical device / SaMD EU MDR · EU AI Act · ISO 42001 · NIST AI RMF
Hospital system / telehealth HIPAA · HITRUST CSF · SOC 2 · NIST AI RMF
Pharma / drug discovery NIST AI RMF · ISO 42001 · EU AI Act (no HIPAA — pharma compound screening doesn't touch patient data)
Healthcare AI vendor (B2B SaaS) HIPAA · ISO 42001 · SOC 2 · NIST AI RMF
EU financial services / fintech DORA · SOC 2 · GDPR · ISO 42001
US federal contractor FedRAMP Moderate · NIST AI RMF
Enterprise SaaS (general) ISO 42001 · SOC 2 · NIST AI RMF
California-operating consumer AI CCPA/CPRA · NIST AI RMF · SOC 2
On cost. ARIA does not absorb your LLM provider's token charges. When ARIA runs a test against your endpoint, those tokens appear on your own provider bill — not ours. A standard evaluation run uses roughly 5–20 LLM calls, costing $0.01–$0.05 on Sonnet-class models. ARIA's platform fee covers orchestration and analysis; the inference bill stays yours.
After the Pilot

What you walk away with.

At the end of the pilot you have real data — not a sales deck, not a demo account. Your own endpoint, tested against the frameworks that apply to your situation, with a live drift threshold running.

Baseline evaluation
Per-fixture pass/fail results across all your active frameworks. This is your starting point — not where you wish you were, but where you actually are. The drift alert threshold is relative to this baseline.
Scheduled re-runs with a live drift threshold
ARIA re-tests your endpoint on the cadence you set. When the pass-rate moves by more than your threshold compared to the rolling baseline, you get an alert. You learn about behavioral shifts before an auditor does.
Optional real-time guardrails
If you activated Shape C, every production LLM call now runs through 7 deterministic pattern gates + 3 AI judges before reaching your users. You changed one URL. ARIA enforces the policy.
Posture Summary PDF
Generate a Posture Summary from ARIA at any time — framework posture, open findings, evidence completeness, and trajectory trend. Shareable with your board, your compliance officer, or an auditor. No re-work required.
A clear path forward
If ARIA surfaces something useful, the same configuration carries forward into an ongoing subscription — nothing needs to be rebuilt. Aggi Technologies will work with you on a pricing proposal that fits your endpoint count, evaluation frequency, and framework requirements.
This is the entry point to ARIA

The pilot connects your endpoint and runs your first evaluation. ARIA — the continuous compliance platform — keeps the work going: monitoring behavioral drift, enforcing policy inline, generating audit-ready evidence across 12 frameworks, and producing the Posture Trajectory your compliance officer and board actually want to see. About ARIA → · Launch Platform

Ready to see where your AI actually stands?

Start with a conversation. Tell us what you're running, where it's deployed, and which frameworks matter to your situation. We'll get you connected and running your first evaluation.

Start the Pilot Conversation → About ARIA
Dr. Prasad N. Golla  ·  prasad@aggicorp.com  ·  (972) 351-3342  ·  aria.aggicorp.com