FAQ
General
What is ATHENA?
ATHENA is Universal Trust Calibration Infrastructure™ — the world's first platform for measuring and optimizing human trust in AI systems. We help organizations detect automation bias, prevent algorithm aversion, and maintain regulatory compliance.
Why do I need trust calibration?
When humans work with AI, they exhibit predictable trust patterns. Some overtrust AI (automation bias), blindly following recommendations. Others undertrust AI (algorithm aversion), ignoring correct suggestions. Both patterns lead to suboptimal decisions and regulatory violations.
The EU AI Act Article 14 requires "effective human oversight" — which means detecting and addressing these trust miscalibrations.
How does ATHENA determine if AI was correct?
Our patent-pending counterfactual analysis determines AI correctness without requiring ground truth labels or waiting for outcomes. This proprietary methodology, validated across 390M+ records, is a core differentiator of the ATHENA platform.
Integration
How long does integration take?
Most customers integrate in under 1 hour:
Install SDK (1 minute)
Add API key (1 minute)
Log first decision (5 minutes)
View dashboard (immediate)
Which languages are supported?
JavaScript/TypeScript —
npm install @athena-ai/sdkPython —
pip install athena-sdk
More SDKs planned for Q1-Q2 2026.
Can I use the API directly?
Yes! All SDK functionality is available via REST API. See the API Reference.
Does ATHENA work with my AI system?
ATHENA is AI-agnostic. We work with any system that makes recommendations and tracks user decisions — from LLMs to traditional ML models to rule-based systems.
Data & Privacy
What data do you collect?
We collect:
User IDs (anonymized)
AI recommendations
User decisions
Outcomes (optional)
Timestamps
We do not collect PII unless explicitly provided.
Where is data stored?
Production: AWS us-east-1
EU Option: AWS eu-west-1 (Enterprise tier)
Database: Supabase (PostgreSQL)
How long is data retained?
Default retention: 2 years (regulatory requirement for audit trails)
Enterprise customers can configure custom retention periods.
Can I delete my data?
Yes. Contact [email protected] or use the API:
Compliance
Which regulations does ATHENA support?
EU AI Act — Article 14 (human oversight)
Texas TRAIGA — HB 2060 (meaningful human review)
Colorado AI Act — SB21-169 (impact assessments)
FDA AI/ML Guidelines — Performance monitoring
California SB-53 — Governance reporting
Are you SOC 2 certified?
SOC 2 Type I is in progress (expected Q2 2026). SOC 2 Type II planned for Q4 2026.
For enterprise customers, we provide:
Security questionnaire responses
Bridge letter
Custom security reviews
Do you sign BAAs for HIPAA?
Yes, for Enterprise tier customers. Contact [email protected].
Pricing
How much does ATHENA cost?
Pilot
Free
100 req/min
Growth
$499/mo
500 req/min
Enterprise
$2,000+/mo
1,000+ req/min
Is there a free trial?
Yes! The Pilot tier is free for your first 90 days. Perfect for integration and initial testing.
What happens if I exceed my rate limit?
Requests return HTTP 429 with a Retry-After header. Our SDKs handle this automatically with exponential backoff.
Technical
What's your uptime SLA?
Growth: 99.5% uptime
Enterprise: 99.9% uptime
Real-time status: status.athenatrust.ai
How fast are API responses?
Average: 45ms
P95: 120ms
P99: 250ms
Can I self-host ATHENA?
Enterprise customers can deploy ATHENA in their own infrastructure. Contact [email protected].
Do you support webhooks?
Yes! Real-time webhooks for:
Trust miscalibration detected
Bias detected
High-risk user identified
Compliance report ready
See Webhooks.
Support
How do I get help?
Documentation
Instant
Self-service
< 24 hours
General questions
Slack
< 4 hours
Enterprise customers
Phone
< 1 hour
Critical issues (Enterprise)
How do I report a bug?
Email [email protected] with:
Request ID (from
X-Request-Idheader)Steps to reproduce
Expected vs. actual behavior
How do I request a feature?
Email [email protected] with your use case. We prioritize based on customer feedback.
Still have questions? Contact [email protected]
Last updated