Engineering drawings
P&IDs, mechanical sets and isometrics, scanned or native.
Backed by Y Combinator
AI that understands the physical world.
Operon turns the drawings, sensor data, documents and email your company runs on into context, then builds AI employees that act on it. Built for your company. Live in three weeks.
The problem
AI employees need context. In industry, it's locked in data no system reads together.
A plant or a project runs on drawings, sensor histories, specs, photos and email. None of it is structured, so AI can't work on it reliably, and your best people spend their days piecing it together.
Vibration, mm/s7.1
Time series
Historian tags, sensor readings and alarms.
23 31 13 · DuctworkDuct in return plenums shall be internally lined.
Documents
Specs, datasheets, procedures and permits.
Inspection · Bay 4Field records
Photos, inspections and work orders.
MH09:41
Marcus Hale
to Estimating
RE: Addendum 2
Revised M-301 attached: two VAV boxes moved and the AHU-3 schedule updated. Bids due Friday.
M-301 Rev 2.pdfLinked to M-301 · 3 changes
RFIs, addenda, quotes and vendor threads.
How it works
From data, to context, to action.
Operon does two things. It builds the context, then it builds the AI employees that work on it.
The context layer
Your data, connected.
Operon reads your drawings, time series, documents and email, and links every tag, line, room and job across them into one model of your company. Every fact keeps a link to where it came from.
Scanned drawings, CAD exports, historian tags, PDFs and inboxes. Nothing to clean up first.
The pump on the P&ID, its vibration trend, its datasheet and its work orders become one record.
Every fact links back to the sheet, reading or message it came from, so your team can check it in seconds.
Context
P-101 · Crude feed pump
Unit 300 · Running
- Drawings3 sheets, latest Rev D
- Vibration7.1 mm/s, rising
- DatasheetAPI 610, single seal
- Work orders12, last in March 2025
- Email2 vendor threads
Every fact links to its source
AI employees
Built for your company.
On that context, we build AI employees for the work your team does every day. Each one follows your company's rules, works in the systems you already use, and sends its work to your people to approve.
Built around how your company estimates, maintains, reviews and reports, not a generic template.
Email, file shares, ERP, the maintenance system and the historian. Nothing to replace.
Nothing leaves without a person's sign-off, and every step is logged.
Northline Mechanical
AI employees · Today
- Done
Bid inbox
38 emails filed to their jobs, 4 replies drafted
- Approve
Take-off · Maple Office, Level 2
Updated for Addendum 2
- Approve
Submittal review · AHU-1
2 deviations from the spec, each cited
- Done
Service agreements
12 renewals due, quotes drafted
Four industries. Each with its own data and its own work.
Pick yours to see what Operon does there.
How we deploy
Live in three weeks, with our engineers inside your company.
We don't hand you software and leave. For the first month our engineers work with your team, connect your data and build your AI employees around how your company already works.
3 weeks
from kickoff to your first AI employees at work
One payment
a single down payment covers the whole cycle, with no additional cost
$2.8M
saved a year, on average, by what we deploy
Week 1
Pick the work
We sit with your team and find the work that costs you the most.
Weeks 1–3
Build
Our engineers connect your data and build your AI employees on it.
Week 3
Go live
Your AI employees start work. Your people approve what they send.
Month 1
Measure
We stay through the first month, tune the work and measure the savings.
Your people bring judgment, experience and the final say.
Operon brings the context, and the hours to act on it.
For private equity
We transform the industrial companies you own.
For firms that buy and run HVAC, mechanical and industrial services companies. We go into each company, connect its data and put AI employees on the work that costs the most. What works at one company is set up at the next.
Each company gets AI employees built for its own data and its own work, live in three weeks.
A workflow that pays off at one company is set up at the next, so each new acquisition starts faster.
What's live at each company, what's being built, and what it saves.
Portfolio
AI rollout
- Live
Platform company · HVAC
6 AI employees at work
- Building
Add-on 01 · Electrical
Week 2 of 3
- Week 1
Add-on 02 · Fire protection
Picking the work
- Next
Add-on 03 · Plumbing
Closes next month
Questions, answered.
AI employees built for your company, on a context layer that connects your drawings, sensor data, documents and email. Each one does a defined job in the systems you already use, and your people approve its work.
Engineering drawings such as P&IDs, mechanical sets and isometrics; historian and sensor data; specs, datasheets and procedures; photos, inspections and work orders; and email. Scans, PDFs, CAD exports and live feeds all work.
No. Making sense of messy data is the work Operon does. Our engineers start with what you have.
One down payment covers the whole cycle, with no additional cost. Our engineers work inside your company for the first month, and your first AI employees go live within three weeks.
HVAC and mechanical contractors, construction and EPC firms, energy operators, and chemical and process manufacturers, including companies owned by private equity.
Each company's data is kept separate, or Operon runs inside your own network. Data is encrypted in transit and at rest, access is role-based, every action is logged, and we never train models on your data.
Operon
Start with the work that costs you the most.
Tell us where your team loses the most hours. Three weeks later, an AI employee built for your company is doing that work.