Agents and workflows
- Define and debug Agents
- Design visual workflows
All products
NexusAI StudioNexusAI
Build AI around the way your team works.
An open-source studio and runtime for creating AI agents, designing workflows and working with enterprise knowledge.
NexusAI StudioProduct responsibilities
Product responsibilities
Agent & workflow studio
01 / Product overview
The work this product is responsible for.
NexusAI Studio brings Agents, workflows, knowledge bases and tools into one environment for AI developers, business teams and platform administrators. Teams can turn enterprise Q&A, document processing and recurring tasks into reusable applications, define how models, knowledge and tools contribute, and refine the results through debugging.
Who it is for
- AI application developers
- Teams building business applications
- Enterprise platform administrators
Inputs & context
Instructions, models, knowledge and tools
Outputs & deliverables
Agent definitions and AI workflows
Its role in the ecosystem
Turn specific tasks into reusable AI applications
NexusAI Studio combines Agents, visual workflows, document knowledge and specialized tools into reusable AI applications. Teams define responsibilities, inspect answers and actions, and maintain applications with clear inputs, processing steps and outputs.
From task requirements to a reusable application
- 01
Define the task and materials
Gather user questions, available materials and expected outputs; establish Agent responsibilities or workflow requirements.
- 02
Build and debug the application
Combine Agents, workflows, knowledge bases and configured tools, then check answers and actions with real tasks.
NexusAI Studio - 03
Deliver and reuse capabilities
Maintain applications and input/output definitions so teams can reuse Agents, knowledge, workflows and Python Skills.
Core capabilities
Knowledge and specialized logic
- Build document knowledge bases
- Package Python Skills
Tool composition and team reuse
- Combine tools and MCP capabilities
- Organize team application assets
Product relationships
Studio builds and runs AI applications. Agent responsibilities and instructions reach Runtime with tasks through OS; Atlas governs connected use, while Terrain and Studio cover business modeling and application creation respectively.
Definition referenceNexusAI OS
Provide Agent definitions to OS
OS references NexusAI Agent settings when assigning AI work, and OpenFlux Runtime runs the tasks and returns the results. Studio creates and configures those Agents.
Capabilities & deliverables involvedAgent responsibilities, instructions and configuration
Definition referenceOpenFlux Runtime
Deliver Agent definitions to Runtime through OS
Studio sets Agent responsibilities and instructions. After process approval, OS delivers them with tasks to Runtime, which uses them to carry out assigned AI work.
Capabilities & deliverables involvedAgent responsibilities and instructions → OS tasks → Runtime
Governance serviceNexusAI Atlas
Manage usage boundaries through Atlas
Within the scope where Atlas governance is enabled, NexusAI application access and model requests follow organizational permissions, policies and quotas, with audit records retained.
Capabilities & deliverables involvedApplication access, model requests and usage records
Complementary responsibilitiesNexusAI Terrain
Business understanding complements application creation
Terrain provides business objects, rules and action definitions. Studio defines Agents, knowledge and workflows. They take responsibility for business understanding and AI application creation respectively.
Capabilities & deliverables involvedBusiness semantics and AI creation responsibilities
02 / Core capabilities
What you can do.
Define and debug Agents
Set an Agent's responsibility, instructions, model and available tools. Use real tasks to inspect its answers and actions, then adjust the configuration to match the team's requirements.
- Responsibilities and model configuration
- Task input and output debugging
Design visual workflows
Express a process through steps, branches, retrieval and tool nodes. Connect input preparation, model decisions and result delivery into an application workflow that can run repeatedly.
- Steps and conditional branches
- Knowledge and tool nodes
Build document knowledge bases
Import documents and organize enterprise knowledge so Agents and workflows can retrieve relevant content. Provide context for enterprise Q&A, document checking and information organization.
- Document import and knowledge organization
- Knowledge retrieval within tasks
Package Python Skills
Write data processing, calculations and business logic as Python Skills. Define their input and output parameters, run them in a sandbox and reuse them in Agents or workflows.
- Custom logic and parameters
- Sandbox execution and reuse
Combine tools and MCP capabilities
Combine platform tools with Workflows and Skills packaged as callable tools. Describe inputs and results using MCP tool definitions so Agents can invoke the configured capabilities.
- Workflow and Skill tool packaging
- MCP definitions and tool invocation
Organize team application assets
Maintain Agents, workflows, knowledge bases and Skills in a team workspace. Reuse existing definitions for similar tasks instead of configuring everything again.
- Team workspaces
- Reusable application and knowledge assets
03 / Use cases
Start with these tasks.
Answer questions about enterprise materials
Organize policies, product descriptions and operational documents into a knowledge base. Set the Agent's responsibility and answer requirements so employees can query the materials.
What you get
A reusable knowledge Q&A application, with knowledge sources and Agent configuration maintained separately.
Process recurring document tasks
Connect input preparation, knowledge retrieval, model processing and template output into a workflow. Give repeated document tasks a consistent processing path.
What you get
A visual workflow with defined inputs, steps and outputs.
Reuse configured tool capabilities
Package calculations or data transformations as Python Skills, or reuse tools already configured on the platform. Make those capabilities available to the Agents that need them.
What you get
A callable, reusable tool capability with its parameter and result definitions.
04 / How it works
From inputs to results.
- 01
Define the task and materials
Choose work that needs to be repeated. Organize its inputs, expected outputs, document knowledge and required tools.
- 02
Create an Agent or workflow
Configure models, instructions and knowledge. Connect the steps, tools or Python Skills required by the task.
- 03
Debug and reuse the application
Check results with real tasks, adjust the application definition and retain reusable configuration and knowledge in the team workspace.
05 / Scope & combinations
A clear scope for use.
Application creation and execution
Studio defines and runs Agents, workflows, knowledge and tools. It is suited to turning specific tasks into AI applications.
Governance and coordination have separate roles
Atlas governs enterprise AI use. OS references NexusAI Agent definitions to coordinate work across departments involving people and Agents.
Workflows describe application processes
Studio workflows arrange steps and tools within an application. OS handles organizational goals, human tasks and coordination of longer projects.
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 productNexusAI StudioAgent & workflow studio