Our vision
An AI workforce for the physical economy.
Industrial companies are short of people who can read their drawings and do the work around them. We are building AI employees that can: first for single jobs, then for whole roles, then across every company an owner brings together.
The physical economy doesn't lack work. It lacks people to do it.
Estimators, project engineers and compliance specialists take years to train, and there are not enough of them. Much of their week goes to reading drawings, checking documents against each other and writing up what they found. That is the work an AI employee can take, if it can read what they read.
01The physical economy
The companies that build, run and maintain everything else.
Mechanical contractors who heat and cool every building. EPCs who build the plants. Operators who run the pipelines, the power and the process units. Most are mid-sized, specialized and stretched thin, and every one of them runs on drawings.
HVAC & Mechanical
Plans, schedules and specs, read for take-off, bids and submittals
Construction & EPC
Bid packages and isometrics, one register from bid to as-built
Energy
P&IDs, LDAR registers and MOC records, reconciled before the audit
Chemicals & Process
Process safety information and PHA records, kept current
02 · Why it hasn't happened
Generic AI reads text. Industry runs on drawings, sensor data and records.
An engineering drawing is symbols, lines and tables that only make sense together, drawn to conventions that change with the owner, the decade and the trade. The readings, datasheets, work orders and email around it live in other systems. A chat assistant can't follow a line across forty sheets, connect it to the sensor on that line, or tell you which revision an answer came from, and in work that carries liability, a confident wrong answer costs more than no answer. An AI employee for industry needs that context first.
03What we're building
One platform, built industry by industry.
Every industry gets the work its own documents hold. Each job runs on the same platform: it reads drawings and documents in your own legend, does the work end to end, and hands it to a person to approve. Read down a column to see one industry, or across a row to see one kind of work.
Estimating
Take-offs and bid checks.
Mechanical take-offs, checked against the schedules and Division 23
Project delivery
Revisions, submittals and turnover.
Submittal reviews, RFIs and turnover
Compliance and safety
MOC, LDAR and process safety.
—
The office and the archive
The inbox, and answers from past work.
Bid invitations, quotes and questions about past jobs
04Across companies
Then the team crosses companies.
Industrial services are being brought together. Private equity firms buy HVAC and mechanical, electrical, plumbing and fire protection, and specialty construction businesses, and help run them as one platform. Each company arrives with its own drawings, estimators, inboxes and habits. The same AI employees can go to work in each of them, starting with the job that costs the most, while each company's data stays its own.
- In each company
- Take-off, email and the workflows only that company runs, starting with the one that costs the most.
- Across the portfolio
- One view of what is running at each company, with each company's data kept apart.
- Company after company
- The ways of working are shared and the data is not, so the next company starts faster than the last.
$2.8M
a year saved on average by what we deploy
6 figures
a year saved by a single tool
3 wks
from kickoff to AI employees at work
05The road
From reading drawings to a workforce.
We started where reading is hardest and mistakes cost the most. Each step builds on what the one before it can read.
01
Read the physical world
P&IDs, mechanical sets, isometrics, specs and records become data, verified by the people who own them.
02
Do single jobs
Take-offs, revision checks, compliance lists and inbox triage, each done end to end and approved by a person.
03
Fill whole roles
AI employees that carry a role's week, from the first email to the signed deliverable, not just one task in it.
04
Staff the portfolio
The same team in every company an owner brings together. The owner decides what is shared; the rest stays private.
06 · What we believe
Four beliefs behind everything we build.
01
Context is the bottleneck.
Industrial companies already know most of what they need. What holds them back is that it's scattered across drawings, data and records that no one can read together.
02
Trust is earned in the details.
When a wrong number means a lost bid or a gap in a safety file, an AI employee has to show its sources and a person has to sign. That's how trust is built, one job at a time.
03
The team compounds.
Every document one AI employee reads, the others can use. Every company that joins makes the next one faster to start.
04
Built for the people who run these companies.
We deploy on live work, not pilots, and measure ourselves by what changes in the bids, reviews and audits.
Field notes
The thinking behind the team
Bring us the work that costs the most.
Send a handful of drawings from a live job, or tell us which workflow eats your team's week. Within a week you'll see an AI employee doing it on your own documents.