Most general contractors are not short on software. They have Procore or Buildertrend for the job, Sage or QuickBooks for the books, Bluebeam for plans and a shared drive full of subcontractor quotes. The trouble is the space between those tools: an estimator retyping scope sheets, a project coordinator chasing certificates of insurance, a controller rebuilding the WIP schedule every month from three different exports.
InTheory acts as your fractional AI and technology lead. We pick one expensive, recurring workflow, measure how it runs today, and lead a scoped 90-day improvement with your team. That might be AI-assisted quote leveling, a cleaner path from site walk to proposal, or simply fixing how your project management and accounting systems share cost codes. Your estimators, PMs and controller still approve every number that leaves the office.
A mid-size GC might see dozens of bid invitations a month on BuildingConnected and win a fraction of them. Each one still needs plans reviewed, scopes leveled and sub quotes compared. When a senior estimator is the bottleneck, the company either passes on good work or submits rushed numbers with gaps that turn into change orders and margin fade later.
Sub quotes, submittals, delivery tickets, lien waivers and pay applications arrive as PDFs and phone photos. Someone opens each one, keys the important fields into Procore, Sage or a spreadsheet, and files it. The work is slow and quiet, so nobody notices the cost until an expired COI or a missing conditional waiver holds up an owner draw.
Owners, architects and lenders expect regular updates, and homeowners on residential jobs expect them weekly. Superintendents carry the real status in their heads and daily logs. Turning that into a clear owner update takes a PM an hour or two per job each week, and when it slips, the phone calls to the office and the jobsite start.
Many contractors only learn whether a job made money after it closes. Committed costs sit in Procore, actuals in accounting, and hours in a timekeeping app like ClockShark or Busybusy. By the time someone reconciles them into a WIP report, the chance to correct a fading job is often gone, and the surety sees swings nobody can explain.
We fix one at a time, starting with the one that costs you most. Each page shows the steps, the tools, what stays human and what we would measure.
Pull scope, exclusions and pricing from sub quotes into your leveling sheet, flag unpriced spec sections and draft qualifications, so estimators spend bid week checking numbers rather than retyping them.
See the workflow →Build negotiated and design-build proposals from site walk notes, the approved estimate and your clarification library, with a principal reviewing scope, price and terms before anything reaches the owner.
See the workflow →Draft weekly owner and lender updates from daily logs, photos, open RFIs and submittals, so the PM reviews and sends in minutes and owners stop calling the superintendent for status.
See the workflow →Extract vendor, job, cost code, amount and retainage from supplier invoices, sub pay apps and lien waivers, match each one to its commitment, and queue it for AP approval.
See the workflow →Track COIs, endorsements, licenses, safety records and closeout documents per sub and per job, with reminders before anything expires and a person deciding who is cleared to mobilize.
See the workflow →Decide which system owns jobs, cost codes, commitments and hours, then connect estimating, Procore or Buildertrend, accounting and timekeeping so data is entered once and drift is caught weekly.
See the workflow →Replace the hand-built monthly WIP spreadsheet with weekly views of backlog, hit rate, projected margin and over/under billings, highlighting fading jobs while there is still time to act.
See the workflow →Log every submitted bid and proposal, prompt the estimator to check in after the decision date, and record outcomes and loss reasons so you learn which work is worth chasing.
See the workflow →How we deliver this: AI automation for small business or fractional CTO leadership, scoped in writing and run over 90 days. See pricing or the sample scope.
We start with what you already pay for. New tools have to earn their place.
Construction data is less regulated than health or legal records, but much of it is sensitive. Bid numbers, sub pricing and owner budgets are competitive information, and payroll on public work falls under certified payroll and prevailing wage rules. We would keep bid data in tools under company accounts, keep sub pricing out of personal AI apps, and leave a person responsible for anything tied to lien rights, insurance or safety records. Colorado's AI Act matters mainly if AI is used in hiring decisions.
General information, not legal advice.
Defaults are illustrative for construction and general contractors (new contracts awarded per month). Change them to your numbers.
a month in staff time, before any lost sales
Illustration only, not a forecast or guarantee. Recovered staff time becomes a cash saving only if spending actually declines or the time goes into paid work. Tool costs are an assumption; we replace every number here with yours before recommending a scope.
Some takeoff tools now automate counts and areas, and they are worth testing on repetitive work. For most GCs, the bigger savings sit around the takeoff: leveling sub quotes, checking scope coverage, drafting proposals and processing paperwork. Your estimators still own quantities and pricing. We would test any tool on your own past bids before relying on it.
Many contractors use a fraction of what their project management platform can do, and its data rarely flows cleanly into accounting. Often the most valuable work is configuration, cleanup and connecting what you already own. We would check that before recommending any new software, and sometimes the recommendation is to stop paying for a tool.
Front Range construction runs most of the year, but spring and summer are the peak months in the field. Many contractors prefer to diagnose in late fall or winter and pilot before spring. We work around your schedule, keep meetings short, and test with real jobs rather than pulling PMs and supers into long workshops.
Plans start at $2,500 a month for advisory, $5,000 a month for lead plus delivery on one workstream, and $7,000 a month for embedded support with team sessions. The initial engagement runs 90 days with a written scope before kickoff. Software subscriptions and major custom builds are budgeted separately, and we would lay out the expected payback on paper first.
It can be, with the right setup. We use business accounts where your data is not used to train public models, limit access by role, and keep sub pricing and owner budgets out of personal logins. Part of the work is a short written policy on which AI tools staff may use, and for what kinds of information.
In 30 minutes we look at one recurring workflow, what it costs you today and whether there is a credible case to improve it. If there is not, we will say so.
30 minutes with Phil or Martin. No technical brief, no pitch deck. Phil or Martin replies personally, usually within one business day.