Product · Prototype in development
AgentIoT Dashboard
AI-powered, fault-tolerant IoT monitoring portal
AgentIoT combines cross-vendor monitoring with AI agent-based fault detection and diagnosis and safeguarded self-healing through defined corrective actions – in one lean platform that runs directly on edge hardware.
Edge
runs locally on ARM and x86 systems, no cloud lock-in
0
actions without approval – humans decide
MQTT
and REST as vendor-neutral device connectivity
DE/EN
bilingual user interface and documentation
The situation
Many devices, many apps – no overview.
IoT environments in buildings and industrial facilities are usually managed through fragmented, vendor-specific dashboards with separate logins. This costs time, complicates troubleshooting and creates dependencies.
On top of that, many embedded devices become unusable not because of hardware defects but because manufacturers discontinue support, updates or management interfaces. The result is avoidable electronic waste. AgentIoT keeps such hardware in operation and thus contributes to the circular economy.
Core features
One system – from detection to resolution.
Vendor-neutral integration
Devices from different manufacturers are mapped into one uniform data model via MQTT and REST adapters. Further protocols can be added.
Continuous monitoring
Readings such as temperature, CO₂ and humidity are validated, stored as time series and shown in one central interface.
AI-assisted diagnosis
An agent-based assistant detects anomalies and presents finding, evidence and recommendation in a traceable way.
Approval-gated self-healing
Corrective actions such as “increase ventilation” are proposed and only executed after explicit approval – every decision is logged.
Asset & firmware management
Device register, firmware compatibility checks and change history provide transparency across the entire life cycle.
Secure access
Encrypted connection (TLS), sign-in, token-protected interfaces and a complete audit trail protect systems and data.
How it works
Example: air quality in a meeting room
01
Measure
The CO₂ sensor reports its values via MQTT. AgentIoT checks origin and plausibility and stores the reading.
02
Detect
If the CO₂ concentration exceeds the configured level, an alarm is raised with severity, reading and threshold.
03
Propose
The AI assistant explains finding and evidence and proposes the action “increase ventilation”.
04
Approve & act
Only after approval by the operator is the command sent to the fan – fully logged.
Live demo
Experience AgentIoT interactively.
The demo simulates a small office building with eight devices. Trigger events and see how AgentIoT validates readings, raises alarms, explains causes and only carries out corrective actions after your approval.
Please note: this live demo is only a simplified sample that illustrates the concept. It differs from the AgentIoT Dashboard, which is currently under development – the product’s features, interface and design differ.
Telemetry
Trigger a scenario
Simulate typical events and watch how AgentIoT detects, explains and only acts after your approval.
AI assistant
- Finding
- All systems operating normally. No anomalies detected.
Alarms
Approval queue Humans decide
Corrective actions are prepared but only sent to the system after your explicit approval. Every decision is recorded in the audit trail.
Device register
| Device | Type | Room | Value | Status |
|---|
Audit trail append-only
01
Choose a scenario
Start e.g. “Meeting starts” and watch the CO₂ curve.
02
Understand the finding
The AI assistant explains finding, evidence and recommendation in a clear structure.
03
Decide
Approve or reject the proposed action in the “Approvals” tab.
04
Trace
The audit trail shows every action completely and with a timestamp.
Unique selling points
Why AgentIoT?
- Edge-native IoT monitoring and management on cost-efficient embedded hardware
- Smart fault detection and autonomous, approval-gated recovery
- One unified view of devices, status and configuration
- Secure remote access in distributed environments
- Continued operation of legacy or unsupported hardware
- Human in the loop: no action without explicit approval, every decision logged
- One software for x86 and ARM – from industrial PCs to single-board computers
- Applicable in building services and industry (industrial IoT)
Sustainability & market
Circular economy by design
Instead of replacing devices once manufacturer support ends, compatible hardware can continue to be monitored and managed with AgentIoT. This extends device life cycles, reduces e-waste and lowers operating costs.
AgentIoT is positioned between complex enterprise platforms and specialised single-purpose tools. A subscription-based model enables scalable use across customer segments.
Technology
Edge-native and lean
- Runs on Raspberry Pi 4 and compatible ARM and x86 systems
- No dependency on external cloud infrastructure – data stays on site
- Device connectivity via MQTT 5 and REST/HTTPS
- Health, readiness and version endpoints for monitored operation
- Scalable for further use cases such as network operations and fleet management
Target groups
Who is AgentIoT for?
- System integrators and home automation providers
- Property managers and facility management
- Technical service providers and maintenance companies
- IoT infrastructures in educational institutions
- Industry, manufacturing and commerce (industrial IoT)
Phase 1 completed
What the prototype already delivers today.
From June to August 2026, the first development phase was fully delivered together with our development partner IoT-AI.Tech (Aschaffenburg) – all milestones were reached and accepted.
Three-node demonstrator
The same software runs on x86 and ARM architectures – containerised or native – in a connected cluster of three nodes.
Circular economy in action
Two of the three nodes are retired LoRaWAN gateways whose manufacturer platform was discontinued. Today they run as active edge nodes – using less than half a gigabyte of memory.
One validation chain for all data
MQTT and REST data pass through the same pipeline: authorisation, schema, device identity, plausibility, storage, alarm rules and audit. Only validated values appear in the cockpit.
Operations cockpit
Overview, assets, telemetry, alarms, approvals, reports, AI assistant and settings in one interface – responsive on desktop, tablet and smartphone.
AI assistant with a clear structure
Every answer follows the pattern finding, evidence, agents involved, next review and approval. The assistant prepares decisions but never executes actions itself.
Firmware compatibility
A documented hardware and firmware catalogue produces reproducible verdicts – each assessment is stored together with its evidence.
Security by design
TLS encryption, sign-in with enforced password change, token-protected write interfaces and a complete audit trail; data protection by design under Art. 25 and 32 GDPR.
Verified quality
Automated acceptance tests for system health, firmware compatibility and MQTT/REST ingest – all passed. Complete documentation is available in German and English.
Next step: by the end of March 2027, AgentIoT will be extended into a functional, technically validated prototype with all core modules and prepared for market entry.
Development status
From prototype to market
The first development phase is complete: architecture, user interaction concept, backend foundation, firmware compatibility check and demonstrator hardware have been implemented and tested.
By the end of March 2027, AgentIoT will be developed into a functional, technically validated prototype that demonstrates all core modules and is prepared for market entry.
Interested in a pilot project? Get in touch.
Posters
AgentIoT Dashboard in the ERDF/JTF Programme NRW.
The project posters (in German) for “AgentIoT Dashboard – AI-powered, fault-tolerant IoT monitoring portal”, co-funded by the European Union and the State of North Rhine-Westphalia.
Funding
Co-funded by the European Union and the State of NRW
The project “AgentIoT Dashboard – AI-powered, fault-tolerant IoT monitoring portal” is funded under the Go-to-Market voucher with resources from the European Regional Development Fund (ERDF) and the State of North Rhine-Westphalia (ERDF/JTF Programme NRW 2021–2027).
Project objective: development of a digital prototype in preparation for market entry – a vendor-neutral IoT monitoring portal that combines continuous monitoring, AI-assisted anomaly detection and diagnosis, approval-gated corrective actions and integrated asset management, and enables existing hardware to remain in use.
Results: a functional, tested prototype with all core modules, available as a demonstration solution and prepared for market entry. Implementation period: 1 June 2026 to 31 March 2027.
More about the programme: www.efre.nrw
Become a pilot partner for AgentIoT.
Do you operate IoT devices, building services or industrial facilities and want to test AgentIoT in a real environment? We look forward to talking to you.