AI Disclaimer

Last updated: 31 March 2026

1. AI-Powered Platform

Acera Labs uses artificial intelligence and machine learning throughout the Platform, including but not limited to:

Bayesian Media Mix Modelling (MMM) — using PyMC and MCMC sampling to estimate channel contributions
Propensity Scoring — gradient boosted trees predicting customer behaviours (purchase, churn, upsell)
Customer Lifetime Value — BG/NBD probabilistic models predicting future customer value
Customer Segmentation — K-means clustering with RFM scoring
Next Best Action — rule-based and ML-powered recommendations per customer
Autonomous Agents — Claude-powered AI agents that analyse data and generate insights
Joey (AI Assistant) — natural language interface powered by Claude

2. Decision Support Only

All AI-generated outputs — including predictions, scores, segments, recommendations, and reports — are provided as decision-support tools only. They are not a substitute for professional judgement, and you should not rely solely on AI outputs to make business decisions.

3. No Guarantee of Accuracy

While we strive for high accuracy, AI models are probabilistic by nature and may produce incorrect, incomplete, or biased results. Model accuracy depends on the quality, quantity, and representativeness of the data you provide.

4. Model Transparency

Every AI-generated output in the Platform is marked with an AI Disclosure Badge that shows:

The model name that generated the output
The data inputs used
Accuracy metrics (where available)
Confidence level
When the model was last trained

You can request human review of any AI-generated output by clicking "Request Review" on the disclosure badge.

5. Autonomous Agents

Acera Labs's autonomous agents operate on an Observe → Plan → Act → Report loop. All agent actions are logged with full reasoning traces. Agents do not take irreversible actions without human approval. You can review and override any agent recommendation.

6. Data and AI

Your data is never used to train models for other tenants
AI models are trained per-tenant using only your data
When using Claude (Anthropic) for agent reasoning, your data is sent via API. Anthropic does not use API data for model training.
All AI processing is logged in the audit trail

7. Bias and Fairness

AI models can reflect biases present in the training data. We recommend reviewing model outputs for fairness, particularly when using propensity scores or customer segmentation for marketing decisions that may disproportionately affect certain groups.

8. Your Responsibility

You are responsible for:

Verifying AI outputs before acting on them
Ensuring your use of AI-generated outputs complies with applicable laws
Reviewing model accuracy metrics and retraining when performance degrades
Reporting any suspected errors or biases to our team

9. Contact

For questions about our use of AI, contact support@aceralabs.com.au.