Register as many devices as you need. Send as much data as they produce. No throttles, no caps.
The bigger your fleet, the better Connhex gets.
Connect over MQTT, HTTP, WebSocket or CoAP. One identity across all of them. Unlimited ingestion. Queryable the moment it lands.
We built the ingestion engine to handle orders of magnitude more than most fleets will ever produce. Drag the sliders to see how your fleet compares.
Need more? Enterprise plans scale horizontally. Capacity grows with your fleet.
Your fleet produces
2.6
billion datapoints/month
Barely noticeable
Connhex platform capacity
Behind the scenes, Connhex scales its architecture to match your fleet's demand, splitting layers, adding capacity, distributing load. You never manage infrastructure.
All services run together on a single instance. The platform scales vertically until your fleet outgrows it.
Behind the scenes, Connhex scales its architecture to match your fleet's demand. As your devices grow from hundreds to tens of thousands, the platform adapts automatically, splitting layers, adding capacity, distributing load.
You never manage infrastructure, provision servers, or re-architect anything. This is what makes "unlimited" possible.
Starting out
All services run together on a single instance. The platform scales vertically until your fleet outgrows it.
Growing
The platform separates storage from processing to handle more data.
Scaling
Each type of device connection gets its own dedicated capacity. Key services are replicated to handle more traffic.
High demand
Multiple clusters work together. Traffic is distributed across independent deployments for maximum throughput and resilience.
Unique credentials per device, Quantum-resistant encryption, Cyber Resilience Act ready.
Admission control
Flash the image, power up, and admit the device when it shows in the queue. A serial that has enrolled before comes back flagged, and approving it revokes what the previous holder was using.
Inventory, connection history and remote access, without a second tool.
Every device carries its model, its labels and its connection state, so a list of thousands narrows to the ones you mean.
Connect and disconnect events are kept per device, and reconnect counts single out the unstable ones.
Support handles a device from the console instead of from a van.
Connect and disconnect events for every device, so a support call starts with when the unit dropped instead of whether it did.
Flapping detected
5 state changes in 58 seconds. Flagged automatically.
Devices are ranked by how often they reconnect inside a window you choose, so a failing modem or a marginal power supply surfaces before the support ticket does.
I'm still impressed by the low latency our applications have. We achieved a 5x speedup over our already best-in-class mobile apps.
CTO, Astrel Group•Read the story
Point your existing updater at a new endpoint and keep your A/B layout, bootloader, signing keys and build pipeline.
Release management
See versions, compatibility, availability and fleet adoption before a release reaches a campaign.
# Your SWUpdate configuration
suricatta : {
@include "/etc/swupdate/connhex.conf"
};
# Written once at enrollment
tenant = "DEFAULT";
id = "line-3-gateway";
url = "https://edge.connhex.com/things/ddi";
sslkey = "/data/connhex/device.key";
sslcert = "/data/connhex/client.crt";
usetokentodwl = true;
Native DDI support
Your signing keys never leave your build. Connhex serves the artifact and records what each device did with it.
Controlled rollout
Avoid fleet-wide failures by giving each wave a share of the devices. Set the failure rate that aborts the campaign, the hours in which devices may install, and read what every unit did with the release.
Nothing happens without explicit authorization.
Every request authorized. Every device authenticated. Every tenant isolated.
Enforcing authorization on every single request (every API call, every device message, every data query) requires an IAM engine that can keep up without adding latency. We built one. It evaluates policies in microseconds, and its performance stays flat no matter how many rules you add.
mTLS device identity
X.509 certificate per device, issued at provisioning. The device proves who it is on every connection.
Encrypted communications
TLS on every hop: device to platform, service to service, platform to database. No unencrypted paths.
Encrypted storage
Data encrypted at rest at the database layer. Credentials and secrets never stored in plaintext.
Secure authentication
Devices authenticate with certificates or unique keys. Users authenticate through the identity service. No shared secrets.
500M+ rules
permission rules evaluated in less than 200 microseconds
Add as many policy rules as you need. Authorization performance stays flat.
The same authorization engine is available to you. Define custom policies for your own application (tenant scoping, role hierarchies, resource-level permissions) and every check runs at the same speed. Your authorization logic rides on the same infrastructure that secures the platform itself.
policy "fleet-manager"
subjects: team:field-ops
actions: read, update
resources: things:*
tenants: tenant-A
✓ evaluated in μs
AI is useful when it can reach reliable, contextual field data and act on it safely. Connhex provides the controlled layer between devices, software and AI agents.
The missing data layer
Raw telemetry needs identity, relationships, history, permissions and safe paths back to devices before software or AI can use it. Connhex makes that foundation part of the platform.
Telemetry stays connected to device models, owners, installations, relationships and lifecycle state.
Fine-grained permissions control what users, applications and AI agents can read and do.
Authorized software can move from questions to rules, workflows and commands back toward the field.
Millions of datapoints. Millisecond answers.
Real-time. Live status of any device, instantly.
Aggregated. Averages, sums, min, max. Across your entire fleet, in one request.
Downsampled. Raw telemetry compressed into rollups. Dashboards stay fast at any scale.
Filtered. By device, time range, metric, or any combination.
An AI assistant built into the platform. Ask questions about your fleet in plain English. Copilot queries your data and returns structured answers instantly.
No query language to learn. No dashboard to navigate. Support teams, field engineers and non-technical operators use the same interface.
Try asking
“What's the average uptime for TempSensor-v2 this quarter?”
“Which sites had humidity above 80% in the last 7 days?”
“Show me devices that went offline more than 3 times yesterday.”
Which pumps in fleet X consumed more than 200 kWh last month?
Ask anything about your fleet...
Anomaly detected: pump-018
Power consumption 15% above fleet average for model TempSensor-v2. Flagged automatically.
Connhex observes the telemetry patterns produced by your fleet, not manually configured thresholds, but the actual statistical behaviour of each device model. A device that deviates from its peers gets flagged automatically.
No training phase. No labeled data. No data science team required. The longer the fleet runs, the sharper the detection gets.
What you can build
Predictive maintenance. Predict that pump #742 needs a filter change in 14 days.
Energy efficiency. Surface the 10% of devices responsible for 40% of the overshoot.
Automated alerts. Trigger notifications when anomaly detection flags a device.
Fleet health scoring. Rank devices or sites by operational health.
With modern tooling and AI assistants, the first stretch is faster than ever. What follows hasn't changed.
The typical build-your-own timeline
Week 1
The easy part
A broker, a REST API, a database. Maybe an LLM writes half of it. It works.
Week 2–4
The real requirements
Authentication. Multi-tenancy. OTA updates. Device provisioning. Certificates. The edge cases pile up.
Month 2–3
The reckoning
Scaling issues. A security audit that sends half of it back to the start. Fifteen services, and nobody remembers why.
Month 3+
The new normal
Every patch, every protocol edge case, every compliance audit: yours to own, forever. The actual product is still waiting.
services to maintain
weeks behind schedule
We have built IoT infrastructure many times over, for real products, under real production pressure. We hit every wall on this timeline: the OTA rollout that bricked a batch of devices in the field, the certificate rotation nobody had planned for, the multi-tenancy retrofit that took longer than the original build. Connhex exists because we decided to solve this once, properly, rather than patch it forever.
An IoT platform looks simple at first. The first version comes together fast. But the complexity underneath compounds: protocol edge cases, security audits, multi-tenancy, scaling walls, until the platform becomes the product, and the actual product stalls.
Building the first version is the easy part. Maintaining, securing, and scaling it for years is where it gets you. That's the trap.
Connhex handles the infrastructure. You keep full control over your product, with modern tools, AI-native workflows, and an API-first architecture designed to be automated, extended, and integrated into the way you already work.
Build with AI
Context7 gives your LLM up-to-date Connhex docs. MCP lets it connect directly to your fleet and debug in real time. The result: a custom IoT app built in hours, not months.
A white-labeled app, an embedded widget, a tailored dashboard: it's an afternoon with an AI coding agent and our docs.
Connect AI to your fleet
Connect Claude, ChatGPT, or any AI assistant to your Connhex account. Your customers can ask questions about their devices in plain language and get real answers, sourced directly from the fleet.
Generate weekly reports, surface hidden trends, or drill into what's actually happening on-site. No dashboards to navigate. No query language to learn.
Model Context Protocol support. Let AI agents interact with your fleet through a standardized interface.
View on GitHubEvery feature is an API. Build with any language, any framework, any workflow.
API referenceFull API specifications published and versioned. Generate clients, types, and mocks automatically.
Manage devices, push firmware, and query data from your terminal or CI/CD pipeline.
View on GitHubReact to events in real time. Connect Connhex to your own backend, alerting, or automation.
Whether you're writing code, prompting an AI, or wiring a CI/CD pipeline. Connhex fits into your workflow, not the other way around.
You didn't start your company to build an IoT platform.
We did. And we'd like to run yours for you.
There's more to explore. Start from wherever you are.
Ready to build?
Want to read first?
Evaluating platforms?
Not technical?
