AI Transparency Report

How We Use Artificial Intelligence

Complete transparency about our AI systems, capabilities, limitations, and safeguards.

Last Updated: December 2, 2025Compliant with Colorado AI Act (SB 24-205)
AI Models We Use
Specific AI systems powering Apex Omnis features

OpenAI GPT-4

Large Language Model

Used for content transformation: analyzing project files and generating intelligence report drafts.

Human review required
No training on your data
May produce errors or hallucinations

Linear Regression

Statistical Model

Used for earnings forecasts: predicting future revenue based on your past 90 days of data.

Transparent methodology
Uses only your historical data
Not financial advice

Third-Party AI Providers

We use API-based AI services from OpenAI and Anthropic. These providers do not retain or train on your data per their API usage policies. We do not operate our own AI data centers.

Accuracy Metrics & Known Limitations
Honest assessment of AI performance and failure modes

Content Transformation Accuracy

~85%
Factual accuracy (with human review)
~70%
First-draft usability
~5%
Hallucination rate

Known Limitations: May misinterpret technical jargon, fabricate details not present in source files, or miss nuanced context. Always review and edit generated content before publication.

Earnings Forecast Accuracy

±25%
Average error margin
30 days
Maximum projection period
90 days
Historical data required

Known Limitations: Cannot predict market changes, client behavior shifts, or external economic factors. Accuracy degrades beyond 30-day window. This is NOT financial advice.

What AI Cannot Do

  • Guarantee accuracy of generated content
  • Predict future earnings with certainty
  • Replace human judgment and expertise
  • Understand context outside provided data
  • Provide legal, financial, or professional advice
Bias Mitigation & Fairness
How we prevent discrimination and ensure fairness

Human Oversight

All AI-generated content requires human review before publication. Users maintain full editorial control and can edit, reject, or override AI suggestions at any time.

Monitoring & Logging

We log AI outputs and monitor for discriminatory patterns based on protected characteristics (race, gender, age, disability, etc.). Suspicious patterns trigger manual review.

User Reporting

Users can report biased or inappropriate AI outputs via our incident reporting system. All reports are reviewed within 48 hours and feed into model improvement.

No Automated Decisions

We do not make consequential decisions (account termination, pricing, access restrictions) based solely on AI. All high-stakes decisions involve human review.

Bias Limitations

Despite mitigation efforts, AI models may inadvertently produce biased outputs due to training data limitations. We cannot guarantee bias-free results. If you encounter discriminatory content, please report it immediately.

Your Data & AI Training
How we protect your data and what we do (and don't do) with it

We Do NOT Train AI Models on Your Data

Your intelligence reports, project files, and earnings data are never used to train AI models without explicit opt-in consent.

  • Third-Party AI Providers: We use API-based services (OpenAI, Anthropic) that do not retain or train on your data per their API policies
  • Your Intelligence Reports: Not used for AI training
  • Your Project Files: Processed for conversion only, not stored for training
  • Earnings Data: Used only for your personal forecasts, not shared with AI providers
  • Internal Analytics: Aggregated, anonymized usage statistics only (no personal content)

Data Retention

Project files uploaded for conversion are processed in-memory and not permanently stored on our servers. Only final intelligence reports (which you control) are saved to your account.

Your Rights

You can delete your intelligence reports, earnings data, and account at any time. Deletion requests are processed within 30 days per GDPR/CCPA requirements.

Regulatory Compliance
Laws and frameworks we follow
Colorado AI Act (SB 24-205)

Compliant with Colorado's comprehensive AI regulation effective 2026. We conduct impact assessments, provide consumer notices, and implement digital discrimination safeguards.

EU AI Act

Our AI systems are classified as "limited risk" under EU AI Act. We provide transparency disclosures and human oversight mechanisms for EU users.

NIST AI Risk Management Framework

We align with NIST AI RMF best practices for governance, mapping, measurement, and management of AI risks.

SEC AI-Washing Prevention

We accurately represent AI capabilities in marketing materials and avoid over-inflated claims per SEC guidance.

Report AI Issues
Encountered biased, inaccurate, or inappropriate AI outputs?

We take AI safety seriously. If you encounter any issues with AI-generated content, please report them so we can investigate and improve our systems.