AI assistant
Ask the plant's record from the AI assistant you already use.
MaintenanceIQ connects the AI assistant your plant already uses, through MCP (Model Context Protocol), to the maintenance record your team keeps in the app. The assistant can ask about work orders, assets, breakdowns and machine readings, inside your organization. Your planner still decides the work.
What it is for
The assistant asks. The record answers.
Maintenance teams are already asking an AI assistant the questions they used to ask a colleague: what is overdue on this line, when did this pump last fail, is that reading near its limit. When the assistant can see the maintenance record, the answer is the plant's own facts. MCP is how MaintenanceIQ gives the assistant that record.
MCP, the Model Context Protocol, is how an assistant asks another system a specific question and receives a specific answer. In MaintenanceIQ those questions land on the same records the API already exposes: work orders, assets, breakdowns and machine readings. The assistant your company already runs is the one that asks. MaintenanceIQ keeps the record.
The connection stays inside your organization, the same boundary as the API. Another customer's assets, work and readings stay in their own organization. When your IT stops the connection, the assistant can no longer ask. The history stays where your team left it.
Questions worth asking before the meeting
| Question | What the assistant reads | What your team still does |
|---|---|---|
| What is open on this machine? | Open work orders on that asset, with priority and criticality | Assign the job and close it in the app |
| When did it last fail, and why? | Breakdowns and the downtime recorded against that asset | Decide whether the next job is a repair or a PM |
| Is this reading near its limit? | The parameter, the warning and critical limits, and the readings already stored | Act on the warning in the planner's list |
| What did we do last time? | The closed work order, the checklist and the notes on that job | Send someone who can use that history |
These are the questions. The machines are yours. The assistant can answer only from records that exist in your organization.
On the floor
Walk into the meeting with the list already asked
Your reliability engineer asks, before the morning meeting, which A-criticality assets have an overdue preventive job and a reading sitting in warning. The assistant asks MaintenanceIQ and comes back with those assets, each one tied to the overdue job and the reading. The engineer walks in with a list. The planner puts the work on the board.
A technician at the machine asks what was done the last time this press stopped. The assistant reads the last breakdown and the work order that closed it, including the checklist and the note the previous technician left. The person in front of the machine still does the job, still adds the photo, and still closes the work order.
A plant head asks, at the end of the day, which assets took the most unplanned downtime this week. The assistant reads the downtime already recorded. That number should match the reliability view. When it does, the assistant is reading the record. When it does not, fix the connection before anyone treats the answer as the plant's number.
Asset Risk stays a score you can open. Risk is condition likelihood times criticality, and every factor links to the reading, the overdue PM or the breakdown behind it. MCP lets the assistant read that score, and the evidence under it, when someone asks which machine to talk about first. The planner still accepts or dismisses any draft work order Asset Risk has opened.
- One assistant, one organization, one kind of question to start
- An answer you can check against the screen a planner already trusts
- The same work order, closed by a person, on the audit log
- Stop the connection when the assistant should no longer ask
What stays true when an assistant is connected
Your organization only
The assistant sees your plant. Another customer's assets, work orders and readings stay in their own organization.
The same records as the API
Work orders, assets, breakdowns and machine readings. A system that needs to create or update those records still uses the API.
People still run the plant
Planning, assignment and closure stay in MaintenanceIQ, with the same roles, factories and departments. The assistant is there to answer.
Asset Risk stays explainable
The score is still condition likelihood times criticality, and every factor still links to its evidence. The assistant can read that evidence. It does not replace it.
For IT
Connect the assistant you already approved
Pick the assistant your company already allows on the network. Connect it to your MaintenanceIQ organization. Ask one question you can check by hand: the open work on one asset, or the last breakdown on one line. Compare the answer with the screen. When they match, let a planner use it for a week.
We do not require a particular assistant product, and we do not ship a connector branded for one. Your IT brings the assistant. We do not publish a public address for the connection. It is set up for your organization, the way a token is issued for an integration.
An assistant answering a person's question is a different job from a system that must create a work order. Use MCP when someone is asking. Use the API when a PLC, an ERP or another system needs to write a reading, open a job, or read the record on a schedule. Both stay inside the organization. Writes from other systems still go through the API, and every write is still on the audit log.
The record the assistant is asking
- Work ordersThe jobs a planner assigns and a technician closes.
- Condition monitoringLimits, alerts and the readings an assistant can be asked about.
- API accessThe token for systems that need to read and write the same records.
- Asset RiskA score from your own data, with every factor linked to its evidence.
- SecurityIsolation, the audit log, and the boundary around the connection.
Frequently asked questions
Does MaintenanceIQ work with an AI assistant?
Yes. Through MCP, the AI assistant your plant already uses can ask work orders, assets, breakdowns and machine readings inside your organization. Your planner still decides the work.
Can the assistant raise or close a work order?
People raise, assign and close work orders in the app. A system that needs to create a job uses the API. The assistant answers from the record.
Is this the same thing as Asset Risk?
Asset Risk scores each asset from your own data: condition likelihood times criticality, with every factor linked to its evidence. MCP is how an assistant reads that score, and the rest of the record, when someone asks.
Which AI product do we need?
The one your company already runs. MaintenanceIQ does not require a particular assistant, and it does not publish a public address for the connection.
Will the answer match the app?
Check one asset before you rely on it. Open work, the last breakdown and a reading should match the screen a planner sees. When they do, the assistant is reading the record.