Building Automation
for Everyone
Install the complete platform on your building in minutes. Your first site is free.
Your building, connected, in about five minutes
Run one command on any Linux host with Docker. The installer prints a sign-in link, you name your site in the browser, and it licenses itself.
The installer prints a sign-in link before it downloads anything. Sign in, name your site, and it licenses itself — no license key, no registry token.
The Normal Framework is an open building automation platform. It integrates with and enhances the BACnet, Modbus, Niagara, and MQTT systems already in your buildings — then schedules them, commands them, and runs your own control programs on the device.
It runs in a container on your hardware, on your network, and pushes data back out over Sparkplug B, TimescaleDB, gRPC, and REST in Brick or Haystack — so your work stays yours and stays portable. One site or a thousand.
Built for portfolios, not just buildings
Most integration tools were designed for one site at a time. Normal was designed for the hundredth building as much as the first.
Every site reports into a single portal with health monitoring, managed backups, and a per-site tunnel that lets you open the on-site console in your browser — no VPN and no inbound firewall rules.
New gateways can license themselves on first boot with zero touch, and can be commissioned over Bluetooth or IP before they have a network at all.
Upgrade your building automation independently rather than putting a band-aid on the problem, with open APIs and a plugin framework that keeps your work portable.
Your buildings, connected to your AI
Normal ships an MCP server, so an AI agent can work with your portfolio directly — list sites, explore layers and unmodeled points, check scan coverage, read live BACnet values, and help write hooks. It runs over the same authenticated tunnel and the same permissions as everything else.
Point-enrichment and AI-assisted modeling are built into the console too, so raw metadata becomes a usable equipment model without hand-tagging every point. The hard part was never the chatbot — it was having a normalized model and a live access path worth pointing one at.
Premium Edge-to-Cloud Model
No 3rd party SaaS subscription to monitor and secure.
Deploy and manage on-premises, via VM, or cloud.
Who Uses Normal?
Unlike other solutions, Normal Framework is more flexible, reliable, and easily deployable for integrators, facility teams, and building automation providers. Normal integrates into the backbone of your smart building portfolio and existing products, normalizing and managing access to data and controls across BAS, IoT, and analytics platforms.
Software developers
- Own your building onboarding experience and wow customers with self-service tools
- Out-of-the-box secure connectivity and data transport with all on-prem systems
- Standardized data models to suit any application
System Integrators
& Contractors
- Deliver a complete supervisory solution — integration, scheduling, and control — across any building portfolio
- Create new revenue streams by offering add-on services like remote service
Owners
and Operators
- Control the integration process and eliminate repeated vendor integrations
- Maintain control and ownership of data assets
- Manage security and access to underlying OT systems
- Hold vendors accountable for product performance with full transparency
OEMs
- Build connections from new IoT services to legacy systems to capture more value
- Add advanced BACnet, Modbus, and Fox integration capabilities to your product.
Data you can act on
Collecting building data is the easy part. Knowing which of it is real, and writing back to equipment without breaking anything, is where integrations actually fail.
Quality flagged continuously
Every point is evaluated as it reports. Stale, dead, intermittent, flatlined, railed, out-of-range, and non-monotonic points are flagged automatically — filterable in the Object Explorer with live counts, and queryable as an indexed field.
Rules you can test first
Build detectors from presets and tag-based selectors, then backtest a rule against your history to see exactly what it would flag before you turn it on. No surprise alert storms.
Writes that relinquish themselves
Commands carry a priority and a lifetime and hand control back when they expire. Nothing is left overridden because someone forgot to release it.
Guardrails on every write
Per-application policy rules control what each integration is allowed to write. Deny by default, allow specific — configured from Settings or straight from the Object Explorer.
Failed reversions get retried
If a device is offline when a command ends, the release is tracked and retried until it lands — surviving restarts. Anything needing attention is visible on the Commands page.
An audit trail for changes
Schedule edits sync bidirectionally with the devices themselves and every change is logged, so you can answer what changed, when, and from where.
.png)
.png)
.webp){kind=logo}
.webp){kind=logo}
.png)
.png)
.png)
