Engineering and architecture
AI training for engineering and architecture teams, inside Claude and ChatGPT.
Your project managers, project engineers and architects practice on progress reports, fee schedules and drawing revisions, 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 project
Where AI earns its keep in engineering and architecture, and where it doesn’t.
The judgment the lessons practice, in the drawings, estimates and schedules a design firm runs on.
Hand it to AI
First drafts and summaries
- A monthly progress report: this month’s work, next month’s, problems and percent complete by task
- An owner’s review comments into a markup list with a status on each
- A transmittal letter with the job’s numbers, county and limits filled in
- A resume’s list of its five most relevant projects for a qualifications package
Check it first
Anything an owner or contractor relies on
- Bid item quantities, which must match in the estimate, the quantity summary sheets and the plans
- Which revision each sheet’s revision schedule shows
- Scope tasks against the fee schedule
- Who on a qualifications form worked on which example project
Keep it with a person
Calls that carry the risk
- Anything a licensed engineer or architect seals
- Reviewing a contractor’s submittals and RFIs (requests for information)
- Adding money to a work authorization without adding scope
- Whether to pursue a project or pass
Try one
Check the AI’s fee note before it reaches the project manager.
AI drafted the note on why a project is over its fee budget this month. Check each line against what you know before it goes to the project manager.

What you’ll see
What each 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
Six requests, one agreement, asked three times.
Which requests are included in basic services? Same agreement, same six requests, three runs. Spot what changes from run to run. What stays the same is consistent, not necessarily right.

Give it the right material
This project’s dates, not every project’s.
A note to one owner explaining why the design is behind schedule. Pick what goes to the AI: what it needs, what helps, and what only adds noise or another owner’s project.

Catch what it added
Catch the line saying a change has no effect before it reaches the contractor.
You gave it two facts about a drawing revision. It wrote a clean note to the contractor, with a line about schedule and cost nobody gave 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 a proposal coordinator can be compared on the same thing.
Checking what comes back: for example
- Project manager: a progress report whose tasks don’t match the work authorization’s fee schedule
- Project engineer: a fire flow note whose residual pressure isn’t the one the model was run with
- Architect: an issue note listing a revision the sheet’s revision schedule doesn’t show
- CAD and BIM technician: a crossing note that misses two pipes the interference check flagged
- Proposal coordinator: a qualifications form whose project matrix lists someone who didn’t work on that project
How it works
From a short form to people who 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 engineering and architecture firms ask
Is there AI training made for engineering and architecture firms?
Skill Tree is AI training for engineering and architecture firms that runs inside Claude and ChatGPT, and on the website. Its lessons for a firm are set in each role’s work: progress reports, fee schedules, review comments, drawing revisions, requests for information and qualifications packages. 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, project engineers, architects, CAD and BIM technicians and proposal coordinators. The habit each lesson tests is the same for every role, so results compare across groups.
Does it teach people to catch claims the AI added about a change?
Skill Tree teaches people to catch what an AI adds that nobody gave it, such as a statement that a change has no effect on schedule or cost, a dimension or a reason. Checking what comes back is one of the habits every role practices, including figures the AI worked out itself.
Do people need to paste drawings or contracts into it?
People don’t paste drawings or contracts into it. They practice on lessons curated for each role and personalized to work like theirs, so they never paste in a project’s. The lessons also teach what never belongs in a chat, such as another owner’s project, fees and contract terms. Skill Tree doesn’t connect to your firm’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 design 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 →