Clea 2.0 · Agentic framework
Clea 2.0 lets AI agents run the fleet
Operators ask in plain language; the framework already understands every device.
Book a demoarrow_forwardWhat an agent can do
Plain-language requests become coordinated fleet operations.
- monitoring
Query telemetry
Ask for device data and get answers from live and historical telemetry.
- device_hub
Organize the fleet
Group devices and check fleet status without opening a dashboard.
- deployed_code
Create apps from a use case
Describe a use case and get the application and its containers.
- system_update
Launch OTA campaigns
Start update campaigns in plain language, on the device groups you choose.
Query telemetry
Ask for device data and get answers from live and historical telemetry.
How it works
Three building blocks make every device understandable to an agent.
- hub
Native MCP server
Clea 2.0 exposes a native Model Context Protocol server, so AI agents can use Clea as a tool to read telemetry, manage fleets and run campaigns.
- auto_awesome
Zero-config onboarding
Services, apps and edge devices describe their own capabilities and interfaces, so Clea understands them without manual mapping.
- schema
Interface and metadata engine
Real-time metadata with visual mapping editors for every IoT interface and trigger. Model once, reuse everywhere.
Secure by design
Agents get no more rights than the people who use them.
- badge
The user’s identity
The agent acts through SSO and OAuth 2.0 with the user’s identity, so it can only do what that user is allowed to do.
- fact_check
Every action traced
Every agent run and tool call is traced and retained for review.
- neurology
Your choice of model
Connect external LLMs through MCP-compatible clients, bring your own model, run a local model on your hardware or use the LLM built into Clea.
An interface that builds itself
One codebase serves every tenant, brand and market.
Self-building interface
Live metadata generates dashboards, widgets and panels; a micro-frontend orchestrator composes every view.
Dynamic Brand Engine
Tenant styling applied at runtime gives each instance its own brand identity, from the same codebase.
Ready for any market
Every screen is translatable, so you can launch in a new country without changing any code.
Studio AI, natively wired
Explore and manage Studio AI workflow projects inside the same interface.
Agentic AI: FAQ
What teams ask about agents in Clea 2.0.
AI agents can read data and act on the fleet through Clea, always within the user’s permissions.
Clea 2.0 exposes a native Model Context Protocol server, so AI agents can use Clea as a tool.
External LLMs through MCP-compatible clients, or the LLM built into Clea.
Query telemetry, organize device groups, check fleet status, create applications and containers from a use case and launch OTA campaigns in plain language.
The agent uses the user’s identity via SSO and OAuth 2.0, so it can only do what that user is allowed to do.
Yes. Every agent run and tool call is traced and retained for review.
No. Bring your own model, run a local model on your hardware, or use a SECO-managed model.
Yes. Studio AI builds agent workflows, and SECO can deliver custom agents and MCP tools as a service.