Introduction to owlONE AI

Meet owlONE: FAU's Custom AI Platform

Welcome to owlONE, FAU's enterprise AI platform powered by Cloudforce's nebulaONE®. owlONE is securely hosted within FAU's Microsoft Azure environment and enables departments, faculty, staff and students to build custom AI agents tailored to their specific needs. 

This guide provides a high-level overview of how owlONE works, including how agents learn, interact with data, and are deployed across the university.

Getting Started

Any FAU student, faculty, or staff member can immediately start building and testing their own personal owlONE agent—no request or approval needed. Simply log in with your FAU account. 

How to Access owlONE:

1. Go to owlONE 
2. Select Explore Agents
3. Click Create under My Agents

Tutorials: 

Watch our tutorial on Canvas, Creating Your First Agent: Introduction to owlONE (5m 28s)

Personal Agent Builder: 

We've created an agent to help you build your own agent! Try the Agent Builder on owlONE now.

 

Core Architecture: How Agents Learn

owlONE enables more than text-based responses by supporting task-driven workflows and integrations with institutional systems. When you build an agent, you can contextually ground its logic using four distinct, multi-source knowledge vectors:

• File-Based Libraries (Retrieval-Augmented Generation - RAG): Upload datasets, course materials, or proprietary papers. The platform utilizes RAG to ensure responses are strictly restricted to your documents, drastically reducing model hallucinations.
• Dynamic Web Grounding: Point your agent to specific public URLs or university web pages to ensure its knowledge reflects real-time operational updates.
• Live API Integrations *: Build custom API Calls directly within the configuration panel. This allows your agent to securely interact with external servers, apps, or databases to fetch live data streams.
• Model Context Protocol (MCP) *:  Connect your agent to deep institutional data warehouses using standardized, secure OAuth-based frameworks without exposing raw database strings.

* Reserved for Official Agents

Agent Lifecycles

To maintain university governance, owlONE utilizes a tiered publishing system:

• Personal Agents: Developed and tested internally by the creator.
• Official Agents: Published for broader use following review through FAU’s IT governance and service catalog processes to ensure security, data compliance, and institutional alignment

Understanding Token Usage & Model Selection

owlONE operates using a token-based consumption model, meaning usage is driven by how much content is processed during each session. While FAU does not currently charge departments for usage, understanding how tokens work will help ensure efficient and scalable use of the platform.

What impacts token usage:

• User prompts and responses (longer conversations = more tokens)
• File uploads and document analysis
• Agent workflows that trigger APIs or multi-step processes
• Model selection (more advanced models consume more tokens per task)
   

Model selection matters: Different models are optimized for different use cases:

Capability Tier	Example Models	Best For
High-Efficiency	gpt-4o-mini, gpt-4.1-mini	High-volume Q&A, chatbots, basic use cases
Standard	gpt-4o, gpt-4.1	General-purpose agents, file analysis
Advanced	gpt-5, reasoning models	Complex research, multi-step workflows

Choosing the right model for your use case helps balance performance, speed, and overall resource consumption.