Our last essay went up in June. We spent the months since building, and the building taught us something we did not expect when we started: the software is no longer where the value is.
We can build nearly anything a construction firm asks for, and we can build it fast. In a few weeks this fall we stood up fourteen working project sites, each on the public bid documents of a different awarded public works project: treatment plants, pump stations, pipelines, and floodwalls across the country. Every one has the full set of drawings and specifications indexed, a submittal register drafted from the specifications, and the project screened for permits. A new one takes about a day. An ERP for a contractor's front office went from a blank page to printing pay applications inside a week. Work that once needed a funded team and a year now needs a person who knows exactly what to ask for.
When building costs that little, the product stops being the scarce thing. The value sits with whoever understands the work well enough to know what should be built, where it connects, and what correct looks like. That is the lesson of our summer, and it changed what we sell.
So we hold our own products loosely. A contractor that already runs a project system and an accounting system it trusts should keep both, and we will connect to them. A contractor with nothing in place for a job can have ours. If a firm needs something that does not exist yet, we will build it. Either way the firm ends up in the same position: its work moving through one place, sorted and checked before a person picks it up. Which software sits underneath is the least important decision in the whole effort.
Here is what we built along the way.
Chalk Line Studio is for the small contractor, the firm where the owner is also the estimator, the project manager, and the bookkeeper. It is that firm's front office: leads, quotes, invoices, and schedule in one place, with an AI agent named Foreman that drafts the quote, writes the invoice, and briefs the owner each morning.
Workhorse and Daybook are for the larger general contractor, the firm with project engineers, superintendents, an owner, and a design team on every job. Workhorse is the record of the project: drawings, specifications, submittals, RFIs, permits, change management, and schedule, with each party signing in to its own view. Daybook is the ERP for that firm's front office. It carries the money from top to bottom, from the budget by cost code through subcontracts, change orders, invoices, and pay applications to the work-in-progress schedule the controller closes the month on.
EntitleHQ and Munivias work the two sides of the permit counter. EntitleHQ tells an applicant which permits a project needs, which agency holds each one, and the order to pursue them in. Munivias gives the reviewing agency its queue.
GeotechAI digitizes boring logs, draws subsurface cross sections, and runs earthwork takeoffs.
Three things came out of that work besides the tools.
We counted the systems a contractor runs and found twenty-one separate functions, from estimating to payroll, with 128 tools in common use across them. No firm wants another login on top of that. A firm wants the systems it already trusts to work together, so we build connections as readily as we build products.
We learned that the record is the hard part. A submittal log that loses one entry is worse than no log at all, because people trusted it. A project record is evidence and has to be kept like evidence, so each Workhorse project has its own database and its own file store.
And we changed our view of where AI pays off first. In June we wrote mostly about engineering firms and the billable hour. An engineer who finishes in half the time has half as much to bill, and that slows adoption for reasons that are entirely rational. A contractor on a lump sum has the opposite arithmetic. Every hour saved in the front office is margin kept. The pricing that holds design back pushes construction forward, and that is where we have pointed the company.
A firm can already buy software. What it lacks is someone who knows its work and knows these models, who will sit down with its people, find where the hours go, and fit the tools to the way the firm already runs. That is what we sell now. The products are how we deliver it.
We spent the summer building the tools. We are spending the fall putting them to use inside firms.