Skip to main content

Projects

A Project is your central workspace. It groups Teams, Workflows, Tools, and documents under a single initiative.

Creating a Project

Navigate to Projects and click New Project. Provide a name and an optional description.

Once created, the project page gives you access to:

  • Teams — the agent groups available for this project.
  • Workflows — the execution pipelines you've defined.
  • Workflow Runs — history of all executions.
  • AI Assistant — a conversational AI scoped to this project's context and documents.

Project Documents

You can upload documents to a project to give the AI assistant additional context. These documents are available during conversations with the project's AI assistant — useful for providing background material, reference data, or domain-specific knowledge.

Long-Term Memory

Each project can have Long-Term Memory enabled. Under the hood, this activates a knowledge graph built on a formal didactic ontology — not a simple vector store, but a structured graph where every piece of content is classified by its epistemological function and connected through typed semantic relations.

When enabled, the platform's Knowledge Curator processes new content asynchronously:

  1. Classification — incoming content is classified into one of 54 knowledge types across four fundamental classes (Orientation, Explanation, Action, Reference).
  2. Relation extraction — semantic connections between knowledge units are discovered and typed (causal, hierarchical, temporal, didactic...).
  3. Graph storage — knowledge units and their relations are stored in a navigable graph that agents can traverse in 11 structured directions.

The project description guides this process — it tells the Curator what topics matter most, what to focus on, and what to ignore. Be specific.

On the project page you'll see a summary of stored knowledge units and top concepts when Long-Term Memory is active.

tip

Write a clear, focused project description before enabling Long-Term Memory. A vague description leads to noisy, unfocused knowledge extraction.

Deep Dive: Knowledge Engine

Long-Term Memory is powered by ORQO's Knowledge Engine — a didactically structured knowledge graph based on the Web-Didaktik ontology (Prof. Dr. Norbert Meder, 2006). This is what sets ORQO apart from standard RAG systems: your agents don't just search — they navigate knowledge through typed semantic paths.

Read the full Knowledge Engine documentation →

AI Assistant

Each project has a built-in AI assistant you can chat with. The assistant has access to the project's documents, tools, and knowledge base, making it useful for exploring your data, asking questions about your workflows, or getting help with configuration.