AI Providers

FlexotiumPOS uses large language models for the AI Copilot, OCR post-processing, forecasting, and workflow automation. You configure which providers are used and how they fall back.

Supported providers

| Provider | Model | Used for | |----------|-------|---------| | Anthropic | Claude 3.7 Sonnet | Copilot conversations, reasoning, long-context analysis | | Google Gemini | Gemini 1.5 Pro | OCR extraction, structured data tasks, high-volume classification |

You need API keys for each provider you want to use. At least one provider must be configured for AI features to work.

Adding your Anthropic API key

  1. Go to console.anthropic.com and create an API key
  2. In FlexotiumPOS, go to Settings → Integrations → AI Providers → Anthropic
  3. Paste your API key and click Save
  4. Click Test — a short prompt is sent and the response time is displayed

Set Anthropic as the primary provider for Copilot if you want the best reasoning quality.

Adding your Google Gemini API key

  1. Go to aistudio.google.com and create an API key
  2. In FlexotiumPOS, go to Settings → Integrations → AI Providers → Google Gemini
  3. Paste your API key and click Save
  4. Click Test

Switching providers

In Settings → AI Providers → Routing, set the primary and fallback provider:

Primary:  Anthropic (Claude)
Fallback: Google Gemini

If the primary provider returns an error or exceeds the timeout, FlexotiumPOS automatically retries the request with the fallback provider. A banner in the Copilot indicates when a fallback is in use.

Policy governance for AI actions

The AI providers can only take actions within the boundaries defined in Settings → AI & Automation → Policies. No AI action (reorder, adjustment suggestion, notification) is executed outside the policy framework, regardless of which provider is active.

Policies are enforced server-side — they are not instructions sent in the prompt, but hard rules in the application layer.

API usage and costs

FlexotiumPOS displays your estimated API usage per provider in Settings → AI Providers → Usage. This is a count of tokens sent/received and an estimate of cost based on published pricing. Actual billing comes from your provider accounts directly — FlexotiumPOS does not mark up API costs.