Models and work context
- Connect models and organize Agents
- Carry project and conversation context forward
All products
OpenFluxOpenFlux
A place for AI in everyday work.
A desktop agent application that brings models, memory, files, browser automation and tools into one workspace.
OpenFluxProduct responsibilities
Product responsibilities
Desktop AI workspace
01 / Product overview
The work this product is responsible for.
OpenFlux is a desktop Agent workspace for employees, developers and individual users. It brings model connections, project materials, conversations, long-term memory and tools into one place. For documents, web actions, code and everyday tasks, users can state a goal, add context and inspect execution results, then continue existing work in the desktop environment.
Who it is for
- Employees handling materials and everyday tasks
- Developers working with code and development tools
- Individuals managing projects and ongoing work
Inputs & context
Conversations, project context, files and tools
Outputs & deliverables
Completed actions, files and reusable context
Its role in the ecosystem
Carry context into everyday and development work on the desktop
OpenFlux brings models, project materials, conversations, long-term memory and tools into a desktop workspace. Users assign tasks, work with files, web pages and code, inspect deliverables and preserve context for the next task.
From work context to inspectable results
- 01
Prepare the workspace
Connect models and required tools, select a project and add files, materials or previous conversations.
- 02
Assign and execute the task
Let Agents use context with file, browser, document or development tools.
OpenFlux - 03
Review and continue
Inspect generated files and action results. Local scheduled work requires the device and runtime services to be available.
Core capabilities
Files and desktop actions
- Work with files, documents and spreadsheets
- Operate the browser and desktop
Extension tools and ongoing work
- Use code and extension tools
- Set goals and schedule work
Product relationships
OpenFlux works independently, accepts collaboration conversations through Router and uses governed model access in Atlas-managed mode. It shares an engine with Runtime; desktop tools, long-term memory and scheduled tasks remain in the desktop product.
Message deliveryOpenFlux Router
Receive team conversations from Router
Router supplies channel conversations, context and assignments. OpenFlux interprets requests, executes them with Agents and tools, then returns results to Router for a reply in the original channel.
Capabilities & deliverables involvedConversations, context and tasks → Agent execution → returned results
Governance serviceNexusAI Atlas
Use the Atlas gateway in managed mode
In Atlas-managed mode, OpenFlux model requests pass through the enterprise model gateway with governed access, policies and quotas. Independent model connections run according to their own configuration.
Capabilities & deliverables involvedModel requests, access policies and quotas in Atlas-managed mode
Complementary responsibilitiesOpenFlux Runtime
A headless endpoint on the same engine
OpenFlux Runtime is built on the OpenFlux engine, has no interface and takes tasks assigned by NexusAI OS. OpenFlux continues to serve personal work at the desktop.
Capabilities & deliverables involvedExecution engine and Agent work capabilities
Independent application
An independent desktop workspace
OpenFlux can work with its own projects, conversations, model connections and tools. It provides desktop execution within the portfolio, with users choosing the connections they need.
Capabilities & deliverables involvedProject materials, conversation context, tools and execution results
02 / Core capabilities
What you can do.
Connect models and organize Agents
Connect the model services you need and organize conversations and tasks across multiple Agents in the workspace, keeping model selection, discussion and the actual work together.
- Multiple model connections
- Conversations and tasks across multiple Agents
Carry project and conversation context forward
Use project materials, long-term memory and imported conversations in new tasks. Reduce repeated background explanations and carry existing content and records into the next piece of work.
- Long-term memory and project materials
- Conversation import and continued work
Work with files, documents and spreadsheets
Let an Agent use file and document tools to process materials, spreadsheets and presentations. Read inputs, generate content and inspect deliverables within the same task.
- Documents, spreadsheets and presentations
- File reading and content generation
Operate the browser and desktop
Use browser and desktop tools to read pages, find information and carry out page actions. Keep research and the actions that follow it within a continuous task.
- Web reading and page actions
- Desktop tool use
Use code and extension tools
Include code, commands, plugins and MCP tools in Agent tasks. Handle files, perform development work and call or extend the tools you have configured.
- Code and command tools
- Plugins and MCP extensions
Set goals and schedule work
Set a goal for work that needs to keep moving, or schedule a task for a particular time or repeated execution. Execution depends on the desktop device and relevant services remaining active.
- Goal mode
- Scheduled tasks
03 / Use cases
Start with these tasks.
Turn source materials into working documents
Provide project files, existing conversations and web information as context. Ask an Agent to organize the materials and produce documents, spreadsheets or presentation content as needed.
What you get
Review materials, edit content and inspect generated files in one workspace.
Develop within a project
Combine project materials and code context with Agents, commands and development tools to address a task. Keep the discussion, actions and results with the project.
What you get
Follow the code work and its outputs through the task.
Keep routine information work moving
Set ongoing work as a goal, or schedule recurring information tasks. Run them as arranged while the device and services are active.
What you get
Work organized around goals or schedules, with context retained for later use.
04 / How it works
From inputs to results.
- 01
Prepare the workspace
Connect models and required tools, select a project and add files, materials or existing conversations to establish task context.
- 02
Describe the task and execute it
Explain the goal and output requirements. Let the Agent use browser, file, document or development tools, adding information when needed.
- 03
Review the result and continue
Inspect execution results and generated files, adjust the content, and use project context and long-term memory for the work that follows.
05 / Scope & combinations
A clear scope for use.
A desktop work surface
Use files, pages, documents and development tools for everyday work. Tasks run in the desktop environment using the configured models and tools.
Connections determine the governance scope
Models can be connected independently. In Atlas-managed mode, model requests pass through the enterprise gateway; Router delivers external collaboration conversations to OpenFlux.
Goals and scheduling
Goal mode organizes ongoing work, while scheduled tasks define execution times. Local tasks need the device to remain awake and the relevant runtime services to be available.
CONTINUE EXPLORING
Place this capability in your own work.
Review the wider architecture, or bring this product into a conversation about your team and systems.
Current productOpenFluxDesktop AI workspace