Construction
AI training for construction teams, inside Claude and ChatGPT.
Your project managers, superintendents and estimators practice on pay estimates, daily reports and bid forms, in lessons written for their roles and played in the assistant they already have open. You see what each team has shown it can do in lessons, and where to help.
AI on a job
Where AI earns its keep in construction, and where it doesn’t.
The judgment the lessons practice, in the paperwork that moves a job’s quantities and money.
Hand it to AI
First drafts and summaries
- A daily report summary from the foreman’s time card and diary: hours, production, weather and notes
- A morning note on yesterday’s quantities and cost against what was expected, by cost code
- A status note on open requests for information (RFIs) and submittals
- Talking points for a toolbox talk
Check it first
Anything someone relies on
- Quantities, which are counted in more than one place and most often fail to agree
- An invoice’s quantities and prices against the purchase order and delivery tickets
- Bid form quantities against the takeoff, and every addendum in the bid
- Equipment hours on time cards against the machine’s meter or tracking hours
Keep it with a person
Calls that carry the risk
- Whether a trench is safe to enter
- Whether an injury is recordable or first aid
- Whether to bid a job, and which quotes to carry
- Whether to protest an order as extra work
Try one
Check the AI’s overrun note before it reaches the controller.
AI drafted the note on why the job is over its cost budget this month. Check each line against what you know before it goes to the controller.

What you’ll see
What each crew and office team has shown it can do, and where to help.
The dashboard comes from what people did in lessons like this one: where each person is in the course, how they are doing on each habit, and which lessons they missed, never their answers.

Spot what changes
One near miss, scored three times, three ways.
The safety team asks AI to score near misses from 1 to 5. Same near miss, asked three times. Spot what changes from run to run. What stays the same is consistent, not necessarily right.

Give it the right material
The log and the spec, not the whole job folder.
A one-page note on late submittals for Thursday’s job meeting. Pick what goes to the AI: what it needs, what helps, and what only adds noise or personal data.

Spot what it added
Catch the reason AI made up before the crew repeats it.
You gave it two lines about a pressure test that moved. It wrote a tidy note, with a reason nobody told it. Would you send it?

Set in each role’s work
One habit, practiced in each role’s week.
Lessons are written for the roles you choose to train. The habit each lesson tests stays the same, so a project manager and an estimator can be compared on the same thing.
Checking what comes back: for example
- Project manager: a pay estimate whose quantities don’t match the inspector’s measurements
- Superintendent: a daily report summary that leaves out the rain, so the delay has no record
- Estimator: a bid checklist that misses an addendum
- Safety manager: an injury log that counts days away from the day of the injury, not the day after
- Equipment manager: an idle-equipment note that misses a rental still billing after its work ended
- Accounts payable clerk: a payment list that pays one invoice twice, resubmitted under a new number
How it works
From a short form to crews that practice every week.
- See sample lessonsFill out the short form. Within 3 business days we send a preview of lessons written for your kind of work. Free, with no commitment.
- Tell us about your teamsThe roles you want to train and the AI tools they use.
- We set it upLessons set in each role’s work, and your dashboard ready to go.
- Your people startInside Claude or ChatGPT, in the chat they already have open, with the website there for anyone who needs it.
Questions construction companies ask
Is there AI training made for construction companies?
Skill Tree is AI training for construction companies that runs inside Claude and ChatGPT, and on the website. Its lessons for a construction company are set in each role’s work: requests for information (RFIs) and submittals, daily reports, pay estimates, bids and toolbox talks. Each lesson asks for a decision and shows right away why it was right or wrong, and leaders see the habits each team shows in lessons, never anyone’s answers.
Which roles does it train?
Lessons are written for each role you choose to train, such as project managers, superintendents, estimators, safety managers, equipment managers and accounts payable clerks. The habit each lesson tests is the same for every role, so results compare across teams.
Can people in the field use it?
People in the field can use it: a lesson takes a few minutes and works in a phone’s browser, where field crews already enter their time and quantities.
Do people need to paste our work into it?
People don’t paste your work into it. They practice on lessons curated for each role and personalized to work like theirs, so they never paste in their own. The lessons also teach what never belongs in a chat, such as passwords and personal details. Skill Tree doesn’t connect to your company’s software. What Skill Tree keeps and admins see
What does it cost?
$9.99 per person a month, or $100 per person a year, by card or invoice.
See lessons written for your construction teams.
Free, with no commitment: fill out the information below and we’ll send you a preview of lessons written for your kind of work within 3 business days.
If you decide to start
- 1. Tell us about your team. The roles to train and the AI tools they use.
- 2. We set it up. Lessons set in each role’s work, and your dashboard ready to go.
- 3. Your people start. Inside Claude or ChatGPT, in the chat they already have open, or on the website.
Already have an organization? Buy seats →