Getting Started Cameras & Video Detection & Recording Industry & Edge Automation & Events Actions Integration & Connectivity Network & Discovery AI & Remote Control MQTT Modbus Pi4J & Raspberry Pi GPIO ZeroMQ System & Administration Comparisons Use Cases Troubleshooting About & Legal
Home / Documentation / AI Modules and Deployment Responsibility
AI Responsibility

AI modules and deployment responsibility

How Banalytics distinguishes optional connectors and demonstrations from the AI system assembled and operated by an Agent owner.


A configurable engineering toolkit

Banalytics provides modular components that technically capable users can assemble into local processing and automation configurations. Banalytics does not inspect, approve or continuously monitor Community Edition configurations and does not know the customer’s cameras, sensor data, model choices, rules, actions or operational purpose.

Optional AI-related modules are examples and connectors. Banalytics does not develop or rebrand the underlying third-party AI models as Banalytics models. A customer-developed module or configuration remains under the customer’s control unless a written pilot, Enterprise or technology-partnership contract expressly assigns Banalytics a development, validation or operational role.

Gemini and YOLO boundaries

  • Gemini connector: the owner supplies a Google API key, selects the model and decides which Agent components are exposed. User messages and selected context are sent directly from the Agent to Google under the owner’s Google relationship and configuration.
  • YOLO showcase: Banalytics includes one demonstration tiny model for evaluation. The owner may replace it with another compatible model, integrate a Python ecosystem or develop models on the owner’s equipment.
  • Other third-party integrations: the same owner-controlled credential model applies to services such as Cesium and Mapbox, and is the intended model for future optional connectors.

Source code for community-facing YOLO integration modules can be requested for review and technology-partnership work. Modules developed for a closed Enterprise delivery are provided only under the applicable contract.

Deployment responsibilities

The owner determines and is responsible for:

  • the intended purpose and legal basis for the configuration;
  • the equipment, data, model, thresholds, rules and actions;
  • accuracy testing, false-positive and false-negative tolerances;
  • human oversight and the ability to stop or override an automated action;
  • notices, consents, DPIAs and other deployment-specific privacy obligations;
  • the terms and data practices of every third-party service selected by the owner;
  • safe operation of customer-written modules and integrations.

No consequential-use validation

Community Edition modules and demonstration models are provided for evaluation and self-managed use without validation for a particular production, safety or legal purpose. They must not be treated as independently verified for biometric identification, employment decisions, access to essential services, law enforcement, safety-critical control or other decisions that may materially affect a person.

A customer that intends such a use must obtain its own legal and technical assessment. Where Banalytics is asked to develop or validate a closed contractual solution, the parties must document the intended purpose, roles, data, model, oversight, testing, acceptance criteria and applicable AI Act obligations in the contract.

Information available to users

Banalytics documentation identifies AI modules, external API boundaries, configurable permissions, model selection and operational cautions. Release documentation describes the versioned integration. Banalytics is formalising a short AI module card for each Banalytics-supplied connector or demonstration, including intended use, external provider, data flow, limitations and recommended oversight.

Questions about a planned AI deployment can be sent to info@banalytics.live.

Last updated: 23 July 2026.