AI & Data Policy
Version 2026-06-25 · Last updated 25 June 2026
How Provixaa applies artificial intelligence, data handling, and human oversight in Provixaa.
This document is provided for product transparency. It is not legal advice. Have qualified counsel review before relying on it in regulated or enterprise contexts.
1. Scope
This AI & Data Policy (2026-06-25) explains how Provixaa ("Provixaa") uses artificial intelligence in the Service, how we protect your data in AI features, and what you should expect from AI outputs.
It supplements our Terms of Service and Privacy Policy.
2. Aria — two intelligence layers
Aria AI Native (blue): answers computed from your governed project data using the same formulas as dashboards, wizard, and reports. No third-party language-model credits are consumed.
Aria Conversations (purple): language-model synthesis for judgment, recovery plans, and tailored advice when deterministic math alone is insufficient. Available on Predictive Intelligence and Aria Intelligence project tiers with a per-project monthly credit pool.
Aria is scoped to each governed project. Program dashboards use built-in rollups only. Out-of-scope questions are declined rather than answered from generic knowledge.
3. Credits and fallbacks
Weighted credits apply to model-assisted operations (predictions narrative, action items, satisfaction insights, executive summaries, and eligible Aria Conversations topics).
When credits are exhausted or a tier does not include Conversations, Provixaa falls back to deterministic engines or Aria AI Native answers. Core governance features continue; AI enhancement pauses until the next monthly reset or plan upgrade.
Super-admin operators may bypass quota gates for support; usage is still logged.
4. Data sent to AI providers
We send only scoped project context required to answer your prompt when Aria Conversations or other model-assisted features run. We do not use your tenant data to train public AI models.
Conversation history, menu intent, and response source (AI Native vs Conversations) may be stored to restore threads and improve abuse prevention.
You must not submit secrets, credentials, or personal data unrelated to delivery governance in AI prompts.
5. Human review required
AI outputs may be incomplete, outdated, or incorrect. They are informational support for operational and margin decisions — not financial, legal, audit, or investment advice.
You are responsible for reviewing AI suggestions before acting on staffing, billing, client commitments, or recovery plans.
6. Questions
Contact admin@bmc360.io for AI-related privacy or policy questions.