A general contractor in Abilene, Texas, runs twenty houses a year. His subcontractors bid $150,000 jobs and he pays them when the owner pays him, which sounds simple enough until you ask the subcontractors how long it actually takes.
He thinks he pays within 30 days of receiving a pay application. His electrician, his plumber, and his framing crew all say it takes 56.
That 26-day gap is not a rounding error. Across the U.S. construction industry, the cost of slow payments hit $273 billion in 2023, roughly 14% of total construction spending. Most of that cost is invisible to the people writing the checks. It shows up downstream, in crew turnover, delayed starts, and corners cut on materials when a sub needs cash to make payroll.
A 26-day perception gap
Billd, a construction finance company that surveys hundreds of contractors annually, documented the disconnect precisely: general contractors believe they pay within 30 days of receiving a pay application, while subcontractors report waiting 56 days on average. The gap isn't malice or laziness but measurement: GCs track from when they submit the check, while subs track from when they submitted the invoice. Both are right about their own number and wrong about the other's.
Meanwhile, 72% of subcontractors now wait more than 30 days past their contract terms to get paid. That share was 49% just one year earlier. And 35% of general contractors admitted to picking and choosing which invoices to pay on time, as though accounts payable were a suggestion rather than a contract obligation. That confession, incidentally, came from the GCs themselves, not from the subs they owe money to.
In residential work, the situation is better than commercial or public. Residential contractors get paid within 30 days about 48% of the time, according to a Levelset survey of 519 U.S. contractors. Public project contractors hit that threshold just 21% of the time. But "better than government work" is a low bar, and even on the residential side, only 12% of construction companies reported always getting paid on time.
Always. Twelve percent.
Running the numbers nobody runs
What does spreadsheet-based financial management actually cost a typical residential contractor? I built a model around a $3 million annual revenue operation, which is squarely mid-market for a builder running 15 to 20 custom or semi-custom homes.
Start with payment delay carrying costs: if your average monthly billings run $250,000 and your subcontractors experience a 26-day delay beyond contract terms (the documented gap between GC perception and sub reality), and you're carrying that float on a line of credit at 9% APR, the math is simple: $250,000 divided by 30 gives you $8,333 per day, times 26 days equals $216,667 in constantly outstanding receivables, times 9% annual interest yields roughly $19,500 per year in pure carrying cost from the perception gap alone. That money doesn't build anything, doesn't frame a wall or pour a slab or hang a door; it services the delay between when work gets done and when checks clear, a delay that neither party fully measures and both parties blame on the other.
Add underbilling. A construction finance analysis by Wiss, a CPA firm specializing in construction, calculated that a $50 million contractor who submits pay applications just five days late carries $600,000 to $800,000 in unnecessary underbillings at any given time. Scale that linearly to a $3 million operation and you get roughly $42,000 sitting unbilled, not because the work wasn't done, but because someone didn't get the paperwork in by the 25th. At 9% on a line of credit, that's another $3,800 per year in interest on money you earned but didn't bother to invoice.
Then disputed invoices, which industry data puts at 1 to 3% of contract value. At $3 million in revenue, that's $30,000 to $90,000 in amounts under dispute annually, and with typical recovery rates between 60 and 70%, you're eating $9,000 to $36,000 in write-offs, roughly $20,000 at the midpoint, which is money that vanished between a spreadsheet formula and a phone call that never happened.
Finally, and this is the overhead that everyone undercounts: if a project manager spends ten hours per week reconciling spreadsheets, chasing approvals, and cross-referencing invoices against change orders (and GCs self-report spending 56 hours per month on payment management alone, up 27% year-over-year), at $45 per hour that's $23,400 per year in labor that produces no construction output.
Total: approximately $66,400 per year, or 2.2% of revenue, evaporating into the friction between spreadsheets, email threads, and the filing cabinet where change orders go to die.
A $15 million bet that builders will stop
On July 7, a San Francisco startup called Agave closed a $15 million Series A led by Accel, with backing from Y Combinator, Khosla Ventures, 8VC, and Zillow founder Spencer Rascoff. The company was founded by veterans of Intuit and QuickBooks, which tells you something about who they think their customer is: not the enterprise GC with a Procore subscription, but the residential contractor running a $2 to $8 million operation out of a truck and an Excel file.
Agave's pitch is AI-powered construction financial management covering budgeting, invoicing, change orders, and lien releases, all the paperwork that currently lives in email threads and spreadsheet tabs with names like "DO NOT DELETE." The software integrates with dozens of legacy construction ERP systems and claims to process more than 80,000 projects representing over $100 billion in materials. Its AI flags invoice errors, reconciles purchase orders against deliveries, and forecasts cash flow, saving project managers an estimated 10 to 20 hours per month.
They are not alone in the market, just the latest and best-funded. Buildern offers AI billing that auto-matches purchase orders, delivery records, and invoices to eliminate mismatches, while Briq runs AI-powered forecasting across multi-project portfolios for contractors juggling a dozen active jobs. A developer writing on Dev.to described building QuickAdmin, a voice-to-invoice system where a contractor says, "Painted 2 bedrooms and 1 hallway, 12 hours of labor plus materials, payment due in 7 days," and the AI structures a professional invoice from the spoken sentence, no typing required. Hardline raised $2 million in pre-seed funding in May for voice-first AI specifically designed for construction sites, where dirty hands and loud equipment make phone screens useless.
By 2025, the contractor accounting software market hit $3.8 billion in 2025 and is projected to reach $7.2 billion by 2033. Construction management SaaS is even larger at $18 billion. But adoption in residential construction lags commercial by years, because enterprise builders run Procore or Sage while your custom home builder runs QuickBooks and a shared Google Sheet that last got a formula update in 2019.
The counterargument nobody wants to say out loud
Here's the uncomfortable truth that no AI invoicing startup will put in their pitch deck: the 56-day payment gap is not primarily a software problem. It is a power problem: general contractors delay paying subcontractors because they can, and owners delay paying GCs because they can. Each layer in the payment chain uses float as a financing tool, borrowing from the people below them by simply not cutting the check. A small sub who sends an AI-automated late-payment notice to a GC who controls their next three contracts isn't optimizing their cash flow so much as risking their livelihood and every relationship they've spent a decade building. The Billd survey found that 43% of subcontractors don't have enough working capital to cover unexpected expenses, and 59% don't seek capital before they need it. These are businesses living paycheck to paycheck, and the paycheck is 56 days late.
AI can optimize billing velocity, catch math errors, and automate the paperwork that eats administrative hours. What it cannot do is change the fundamental incentive structure that rewards slow payment and punishes the people with the least bargaining power. Subs who accounted for working capital costs in their bids earned 24% profit margins, according to Billd, compared to 17% for those who didn't. The tools to calculate working capital costs are not hard. The willingness to price them into a competitive bid, when the GC across town is underbidding on thinner margins, is the real constraint.
What a $200-a-month subscription actually buys
Back to the math. A typical AI construction finance tool costs $200 to $400 per month, which at the midpoint is $3,600 annually. Against a $66,400 annual spreadsheet tax, even a 30% reduction delivers $19,900 in recovered value on a $3,600 investment, a 5.5-to-1 return that most builders will never calculate because the baseline cost is invisible.
More concretely, if the tool cuts administrative hours from ten per week to six (the 10-to-20-hour monthly savings that Agave claims, applied conservatively), that alone recovers $9,360 per year in project manager time. If automated invoice matching reduces disputed amounts by half, that's another $10,000. If billing reminders and better pay-app timing shave five days off the payment cycle, the carrying cost drops by $3,750.
Nobody in residential construction is making this calculation because nobody in residential construction is measuring the cost of the current system. The spreadsheet doesn't know what it's costing you. That's the whole point.