AI decision engine · physical assets

Know what's failing before it fails.

The decision layer for every physical asset — health, risk, root cause, and the action to take. Built from the documents you already have, not a year of sensor rollout.

CHILLER-2 · TOWER A DECISION · LIVE
31 CRITICAL
Health score
Failure risk78%
Life remaining~40 days
Root causeRefrigerant undercharge · 0.83
Dispatch now — refrigerant-certified tech ACT
New York pilot → U.S. scale-up Brick · Haystack · OPC UA ▲ Memory-native
What it produces

Four decisions, per asset

Insmart doesn't hand you a dashboard of raw readings. It reads the documents a facility already has — manuals, service records, BAS trends, work orders — and turns them into four decisions a facility manager can act on.

AssetEventMaint.
Three inputs in

Whatever the source — a spreadsheet, a PDF, a BAS feed — every asset collapses into three objects: Asset, Event, and Maintenance. The engine reasons over that one model and returns the four decisions below.

01

Health Score

A single 0–100 read on asset condition, updated as new documents and readings arrive.

02

Failure Risk

The probability and rough timeline of failure, so the team can prioritize what actually needs attention.

03

Root Cause

Why an asset is degrading — not just that it is — traced back to the evidence in its history.

04

Recommended Action

What to do next, and who should do it — routed straight into dispatch when a fix is needed.

The loop

Knowledge → decisions → maintenance → insights

Usable decisions in days, from documents — not a year of sensor rollout.

01 · KNOWLEDGE

Ingest

Manuals, service history, BAS points, and work orders are read and structured into an asset-level knowledge base.

02 · DECISIONS

Decide

The Decision Engine produces Health Score, Failure Risk, Root Cause, and Recommended Action for each asset.

03 · MAINTENANCE

Act

Recommended actions route into dispatch — to your own team or an InsmartFix contractor — with the evidence attached.

04 · INSIGHTS

Learn

Every outcome writes back to the knowledge base, so the next decision on that asset is better informed than the last.

KDMI
The data flywheel

Every cycle makes the next one smarter

Knowledge feeds decisions, decisions drive maintenance, maintenance produces new evidence — and all of it writes back to the knowledge base. The wheel turns on its own, and each turn deposits another ring of data the AI learns from. The longer a facility runs on Insmart, the harder its advantage is to replicate.

The loop compounds. Every work order, every technician note, every completed repair becomes evidence the system reasons from next time.

Operations, reimagined

From reactive repair to decision-driven operations

Traditional maintenance waits for things to break. Insmart rebuilds the operations flow as one automated, decision-driven chain — sense, decide, dispatch, learn — each stage feeding the next.

Sense

Asset sensing

Documents, BAS points and events map into one unified asset model.

Decide

Engine decides

Health Score, risk, root cause and recommended action — produced in one pass.

Dispatch

Smart dispatch

Routed to the right technician with the pre-diagnostic already attached.

Learn

Write-back learning

Completed outcomes write back to the knowledge base, so decisions keep improving.

Why it compounds

Memory is the moat

Insmart keeps an evidence-linked memory for every asset — a running record of what was read, what was concluded, and why. Every decision traces back to the documents and events that produced it.

That auditable "why" is what makes a Root Cause call trustworthy, and what makes the next decision on that asset sharper than the last. The knowledge base is the product's durable advantage — it gets harder to replicate the longer a facility runs on it.

ASSET Chiller-2, Tower A — installed 2016, last major service Mar 2025.
EVIDENCE Service log flags refrigerant top-off twice in 6 months — a recurring-leak signature, not a one-off.
DECISION Root cause set to refrigerant undercharge (confidence 0.83), linked to both service records.
OUTCOME Technician confirms leak at brazed joint. Finding written back to the asset's memory.
One engine · three ways to deploy

Use it standalone, or embed it in the systems you already run

Insmart is a decision layer that sits on top — not another system to rip and replace. Whether you start from scratch or already run a BMS, CMMS, or property-management platform, the same engine plugs in at the depth that fits.

Standalone portal

Log in and go

Sign in at insmartai.com, import the documents you already have, and go live in days. No IT project, no vendor coordination required.

Best for: small-to-mid facilities, single buildings, contractors
Enterprise SSO / API

Write back to your workbench

Single sign-on and API access push decisions back into the property team's own workbench, with permission isolation and audit trails.

Best for: large property groups, compliance-sensitive sites
▲

Integration depth is your choice, not a product limit. All three share the same Decision Engine and the same asset model — start on the standalone portal and upgrade to embedded or enterprise integration later, with no migration.

Where it goes next

Built on HVAC, built for more

Phase 1 is HVAC, document-driven, New York pilot. The decision layer generalizes to any physical asset with a maintenance history — these are the industries on the roadmap.

Commercial real estate

Phase 1 · HVAC pilot

Data centers

Roadmap

Cold chain

Roadmap

Healthcare facilities

Roadmap

Logistics

Roadmap

Pools

Roadmap
For contractors · InsmartFix

Every dispatch arrives pre-diagnosed.

InsmartFix routes Insmart's structured work orders straight to your team — asset history, root cause, and recommended parts attached before your technician leaves the shop.

Explore the contractor portal
Voice intake
Smart dispatch
Pre-diagnostic sharing
Workflow optimization
Get started

See your assets' decisions, not just their readings.

Book a walkthrough with the team, or reach out if you're an investor looking to learn more about the pilot.