AI & Automation Overview
FlexotiumPOS integrates AI at the operational layer — not as a bolt-on chatbot, but as a system that reads your live data, surfaces insights, and can take actions within approved boundaries.
AI providers
FlexotiumPOS supports two AI providers:
| Provider | Best for | |----------|---------| | Anthropic (Claude) | Nuanced reasoning, long-context analysis, copilot conversations | | Google Gemini | High-throughput classification, OCR post-processing, structured extraction |
You can configure the primary and fallback provider under AI Providers. Both providers operate under the same policy governance framework — the AI cannot take actions beyond what you explicitly permit.
Policy governance
All AI-initiated actions pass through a policy layer before execution. Policies define:
- What actions the AI can take autonomously (e.g.
can_suggest_reorder: true) - What actions require human approval (e.g.
require_approval_for_adjustments_over: 10) - What actions are always blocked (e.g.
can_delete_products: false)
Policies are configured in Settings → AI & Automation → Policies and require Owner-level access to change.
What's automated vs. what requires approval
| Feature | Mode | |---------|------| | Reorder suggestions | Suggestion only — human approves | | Low-stock WhatsApp alerts | Autonomous | | Forecast generation | Autonomous | | Playbook step recommendations | Suggestion only | | Invoice OCR extraction | Autonomous (human reviews output) | | Stock adjustments from AI | Always requires approval | | Sending WhatsApp messages | Autonomous (within templates) |
Key pages
- AI Copilot — ask questions about your business
- Forecasting — demand, stock, and purchase forecasts
- Playbooks — guided operational procedures
- Automation Triggers — rule-based event automation
- Knowledge — what the AI learns about your business