Kilo IoT Platform

AI-first IoT Platform with an AI assistant that works like an experienced IoT integrator for your devices and data.

Kilo IoT is an AI-first IoT platform for device management, data collection, real-time processing, visualization, automation, and alerting. It is available as a fully managed cloud deployment or as a self-hosted on-premise installation.

What makes it AI-first is AIoT — Artificial Intelligence of Things built into the core of the server, not bolted on as a chat box. A built-in AI assistant works like having an experienced IoT integrator by your side: ask it in plain language to onboard a device, build and deploy an automation, or stand up an alarm, and it does the work with you — grounded in your live deployment and always under your confirmation.

Kilo is built for organizations that need to connect, monitor, and automate physical devices at scale — whether that means five sensors in a single building or thousands of devices across dozens of sites. When we designed the server, we chose full customizability over a simpler but limited tool — and AIoT is what makes that depth easy to use: the assistant turns a highly configurable platform into something you operate by describing what you want.

Kilo can also operate as a Physical AI platform for external AI models and agents. It places consistent device state, permissions, tested automation, commands, verification, and operational history between model intent and real assets. Read Physical AI for AI Agents for the architecture and integration paths.

Who Kilo IoT is for

Kilo IoT is designed to serve any IoT use case where devices need to be connected, monitored, and automated — regardless of industry or vertical. The server provides the infrastructure; you define what it does.

Organizations across a wide range of sectors use Kilo to solve fundamentally similar problems: collecting data from distributed devices, making that data actionable, and automating responses when conditions change. Some examples:

• Operations teams managing environmental sensors, equipment monitors, or utility meters across facilities

• System integrators deploying IoT solutions for clients who need a reliable, multi-tenant server

• IT departments looking to centralize device data without building a platform from scratch

• Facility managers tracking conditions across buildings, floors, and zones

• Agricultural operations monitoring soil, weather, and irrigation systems

• Logistics providers tracking assets, cold chain compliance, and fleet conditions

• Smart building operators managing HVAC, energy, occupancy, and access

These are examples, not boundaries. If your use case involves connecting devices and acting on their data, Kilo is built for it.

What the platform does

AIoT — an AI assistant that works like an IoT integrator

The built-in AI assistant is not a generic chatbot pinned to a help page. It is connected to your actual deployment and works like an experienced IoT integrator who knows your system end to end. Open AI Chat from the sidebar and brief it the way you would a colleague.

It answers from your real data. Ask "What was the average temperature in Warehouse B last week?" or "Which devices haven't reported in 24 hours?" and it queries your device history, analyzes trends, and gives a grounded answer — scoped to your permissions, never invented.

It does the work, not just the talking. Describe an automation in plain language and it designs the rule, writes the CEL, tests it, and deploys it. Ask it to onboard a device and it runs the flow; ask for an alarm and it builds the escalation chain. Before anything consequential, it pauses for your explicit confirmation, then reads the result back to verify its own work.

This runs on an enterprise-grade agent runtime built in-house and improved continuously — its accuracy grows as the agents are trained on more real-world IoT work, so the assistant keeps taking on more. That runtime is Synthetic Brew, built as a product in its own right rather than a chatbot wrapped around someone else's API.

Device management with Digital Twin technology

Every device you register becomes a full digital representation — a Digital Twin — that mirrors its state, configuration, telemetry history, and behavior in real time. The digital twin persists even when the physical device goes offline, so you always have a complete picture of your deployment.

You are not limited to a pre-approved device catalog or a single protocol. Connect LoRaWAN sensors, MIOTY endpoints, vehicle trackers, and any MQTT-capable device — PLCs, energy meters, building systems, and custom-firmware hardware all connect through the appropriate connector. Different manufacturers report data differently — one vendor sends temp_c, another uses temperature, a third just sends a raw number. Kilo normalizes all of it through sensor templates, normalized key names, and unit measurements. You define how each device's raw data maps to standardized parameters, and the server handles the rest. Once mapped, every device — regardless of manufacturer or connectivity method — reports data in a consistent, unified format.

For many popular devices, premade templates with pre-configured mappings are already available — simplifying onboarding to just entering device credentials. For any other device, you map its raw output to meaningful parameters yourself. There is no dependency on a pre-approved device catalog.

Built-in LoRaWAN Network Server

The server includes a fully integrated LoRaWAN Network Server (LNS). There is no need to deploy, configure, or maintain a separate LNS. Device join requests, uplinks, downlinks, and message deduplication are handled automatically. This removes an entire layer of infrastructure from your deployment.

Built-in MIOTY Service Center

Kilo Cloud includes the Enterprise edition of KiloCenter, Kilo's MIOTY service center. You do not deploy or integrate a separate network server: add the MIOTY connector, connect the base stations, and register the endpoints. Their readings enter the same normalized device model, dashboards, rules, alarms, audit history, and AI workflows as data from other Kilo connectors.

If your requirement ends at the MIOTY network layer and you want to operate it yourself, KiloCenter Community Edition is free, open source, and self-hosted. Use Kilo Cloud when you also need the complete IoT operations platform around the MIOTY data or want the Enterprise service center managed for you.

Visual rules engine with production-grade safety

Automation rules are built on a visual canvas using a drag-and-drop editor. Conditions are written in CEL (Common Expression Language) — a fast, safe expression language that goes far beyond simple threshold comparisons. A rule can be as simple as device.temperature > 30 or as complex as your operation requires — combining multiple sensor readings, spanning devices across different locations, evaluating decision tables with multi-criteria hit policies, and branching through conditional switch logic. There is no artificial ceiling on complexity.

Every rule change is automatically versioned. You can review the full history of any rule, compare versions, roll back to a previous state with one click, and clone rules as templates. Rules are built into deployable artifacts and deployed to production explicitly — giving you full control over what's running and when. This means automation you can trust in production — not automation you have to babysit.

Rules also support decision tables for structured multi-criteria evaluation and schedules with day-of-week and timezone selection. When a condition must persist before the rule starts, create a trigger and choose Only if it lasts. The trigger owns that duration and can watch one or several devices independently; the rule's Start Event chooses the trigger and optionally restricts execution to scheduled hours.

Multi-channel alerting with escalation

When an alarm rule fires, the system delivers notifications across email, SMS, and push simultaneously. Each channel is independently configurable per rule. Alarms are categorized across five severity levels — Critical, High, Medium, Low, and Info — each with its own repeat interval and escalation behavior. Organizations can configure per-severity policies to control how each level is handled across the deployment.

Escalation steps let you define notification chains: if an alert isn't acknowledged within a set time, it escalates to the next recipient or channel. Quiet hours and weekly scheduling windows ensure that non-critical notifications respect operational schedules.

The notification delivery system is fault-tolerant. If a notification fails to send — because of a temporary email provider issue or an SMS gateway timeout — the system automatically retries with exponential backoff. If the automation service restarts during execution, workflows resume from where they left off. No alert is silently dropped.

Custom dashboards and real-time visualization

Build operational dashboards organized in a folder hierarchy — by site, building, floor, or any structure that reflects your deployment. Each dashboard holds widgets sourced from any device and any parameter.

Widgets are fully customizable. Set value boundaries to define normal operating ranges, choose alternative units, toggle graph visibility, and switch between display modes. Each widget type — line charts for trending data, numeric displays for live readings, boolean indicators for on/off states — can be configured to match your operational context. All data updates in real time.

The platform's transport infrastructure is protocol-agnostic — data from LoRaWAN, MQTT, and vehicle tracker devices all enters through the same pipeline, gets standardized and normalized, then flows through specialized channels to dashboards, the rules engine, and historical storage simultaneously. Live data reaches dashboards and automation rules through real-time streaming — no polling, no refresh cycles. Real-time streams are also available externally via WebSocket and Server-Sent Events (SSE).

The entire pipeline — from message queues to caching layers to query storage — uses specialized tools and an architecture designed for speed and scale:

• Historical queries across large datasets are optimized for fast response times

• Dashboard charts are designed to load quickly, even when visualizing months of data

• The architecture is built to scale as your deployment grows

• Data is partitioned by time for efficient retention and compressed for long-term storage

Digital Building Twin — your facility modeled in 3D

The Digital Building Twin is a 3D model of a real facility, built directly into a dashboard. Draw the building yourself — walls, doors, windows, multiple floors — or import an existing CAD floor plan, or trace the outline from an aerial map. Furnish it from a catalog of more than 60 ready-made 3D objects: desks, racks, parking spots, AC units, pumps, vehicles, gates. Then bind any object to a sensor in your deployment and define what its colors mean.

Once it is wired, the model becomes a live operational view. A cold store shades amber as it drifts out of range. An occupied parking bay turns red while a free one stays clear. A pump turns red the moment it faults. Instead of mapping a row in a table to a place in their head, an operator looks at the building and sees exactly where attention is needed. Occupancy, parking availability, equipment health, climate faults, zone conditions — whatever your sensors report, it shows up where it physically belongs.

The building can also be anchored to real GPS coordinates, giving the model a true position in the world. See Digital Building Twin for the full guide.

Maps, GPS tracking, and spatial views

Place devices on 2D maps for spatial context. For GPS-equipped trackers, review full location history with playback, coordinates tables, and route visualization.

Attribute-Based Access Control and multi-organization isolation

Kilo uses Attribute-Based Access Control (ABAC) rather than traditional role-based access. Permissions are evaluated dynamically based on organization membership, page-level assignments, and user context — giving you fine-grained, per-resource control without role explosion.

Each organization on the server is fully isolated: its own devices, users, dashboards, rules, billing, and audit trail. A single user can belong to multiple organizations with different permission levels in each. This makes Kilo a natural fit for managed service providers, facility management companies, and multi-tenant deployments.

Audit trail

All membership and access events within an organization are logged — user invitations, acceptances, permission changes, and removals. The audit trail provides traceability for compliance and operational accountability.

API access and integrations

Kilo IoT Server exposes a public API for backend integrations, analytics and reporting pipelines, industrial and automation tooling, and custom applications. REST is the primary path for standard cloud and HTTP integrations; a gRPC interface is also available as the advanced / on-premise path for typed clients and service-to-service integration — choose REST unless you specifically need gRPC. Requests use scoped API keys (X-API-Key) that you can limit to specific permissions, expire, and rotate. Real-time data is also available via Server-Sent Events (SSE) and WebSocket streams. See the API section for protocols, authentication, and examples, and API Keys to create keys.

Deployment options

• Kilo Cloud — Fully managed, scalable, and fault-tolerant. No infrastructure to maintain.

• Kilo On-Premise — The same server deployed within your own infrastructure, for organizations that require full control over data, access policies, and network boundaries.

Languages and accessibility

The interface is available in English, German, French, Spanish, and Portuguese. Both light and dark themes are supported, and your preferences persist across sessions.

Subscription plans

Kilo IoT offers multiple plan tiers — from a free tier for evaluation up to enterprise plans for large-scale deployments. Each tier defines limits for devices, automation rules, and access to advanced features. Plans can be viewed and managed from the Subscription section in the user menu.

Access the Platform

Launch the Kilo IoT Server at app.kiloiot.io.

Where to start

If you're new to Kilo IoT, begin with the Getting Started guide — it walks you through the interface, explains what makes the server different, and takes you from zero to a working deployment with live data and alerts.