AI Driven Implementations
AI-Driven Implementations#
ARPIA offers three ways to apply AI to your data. All three run inside the platform's access controls, and you can combine them in a single solution.
Choose an implementation#
| If you need to… | Use | What it does |
|---|---|---|
| Ask questions about your data in natural language | AI Agent | Interprets the question, queries authorized data, and returns an answer, table, or visualization |
| Get answers grounded in documents and knowledge sources | RAG AI | Retrieves relevant content at query time and generates a response from it |
| Run a repeatable process on your data | AI Worker | Executes a defined workflow and produces an output |
Example uses#
- Self-service analytics: business users query governed data without SQL (AI Agent).
- Policy and procedure lookup: answers that cite internal documentation (RAG AI).
- Compliance checks and anomaly detection: scheduled or triggered workflows that flag exceptions (AI Worker).
Models#
The implementations use ARPIA's internally designed models and third-party models through the multi-provider LLM gateway. See Model Catalog.
Governance (applies to all three)#
- Role-based access. An implementation can only reach data the user's Security Profile allows.
- Read-only by default. Write access requires explicit configuration. [VERIFY for AI Workers, which produce "operational updates"]
- Scoped by workflow. Each implementation can only do what its configured workflow defines.
- Configurable human review. You decide where a person approves or reviews output.
- Activity logging. See Logs & Usage. [VERIFY: what is logged]
- Change management. Changes to AI features follow a structured lifecycle and change process.
For policies, risk, and monitoring, see AI Governance.
Limits#
- Output quality depends on the underlying data and how the workflow is configured.
- Review AI-generated output before you use it for high-stakes decisions.