From QuickBooks data to what to change, and what it’s worth.

Click “Connect to QuickBooks” and you’re sent to Intuit’s official authorization screen, where you log in the way you always do and approve access for JobProfitAI. It takes about two minutes.

This uses Intuit’s standard OAuth connection, the same mechanism every QuickBooks app uses. Nothing about your QuickBooks credentials passes through our servers; we receive an access token, which we encrypt before storing.

JobProfitAI reads QuickBooks Projects (sub-customers), your customers if you make one customer per job, or your Classes if each job is a class, and pulls together everything attached to each one: invoices, sales receipts, credit memos, bills, expenses, vendor credits, time entries and estimates. Sales tax is taken out of revenue, refunds and credits come off, and labor from timesheets is costed at the cost rate on each time entry (the employee’s cost rate in QuickBooks), not the rate you bill. If your cost rates are just wages, you add a labor burden for payroll taxes, workers’ comp and benefits in Settings; if they already include those, you leave it at 0. JobProfitAI works with QuickBooks Online only, not QuickBooks Desktop.

Costs are grouped into categories you’d actually recognize: labor, materials, subcontractors, equipment and overhead. Where a cost was tagged to a customer rather than to the specific job, JobProfitAI attributes it only when it can do so defensibly, and flags that it made that judgment rather than hiding it.

The lines on your QuickBooks estimates are sorted the same way, by their product or service, so JobProfitAI knows how each price was split between labor, materials, subs and equipment. If an account or product lands in the wrong category, you move it in Settings.

You set a target margin, company-wide or per job type, and confirm a job type for each job. QuickBooks has no field for job type, so JobProfitAI suggests one from each job’s name and estimate (and, if you ask, from AI reading the job and customer names), and nothing is applied until you accept it. Use the built-in types, rename them, or add your own, such as Kitchen remodel or Service call.

Your Profit Dashboard shows revenue, cost, gross profit and margin for every job, plus company-wide totals: margin across all those jobs, how many are below your target, and the dollar value of profit currently at risk.

Open any job to see the full cost breakdown by category, the estimate-versus-actual comparison, and a profit leakage view showing where expected profit turned into actual profit. One click puts the job’s revenue and costs beside QuickBooks’ own Profit and Loss for the same project, and explains any difference.

The WIP report shows every open job’s contract, cost to date, percent complete and whether you’re over or under billed. The bank-ready version lays it out the way banks and bonding companies read it, with estimated gross profit, cost to complete, totals and the contracts you finished in the last 12 months, ready to print or save as a PDF. It also exports to a spreadsheet for your accountant.

“Needs Your Attention” is a deterministic rule set that runs across every job: costs running over estimate, finished jobs below your target margin, work done but not yet billed, and margin declining across recent weekly snapshots. On Pro, in-progress jobs are also judged on where they’re forecast to finish, and each job’s cost categories are compared with similar completed jobs to catch outliers.

JobProfitAI syncs every night. When that sync finds an open job whose costs are more than 10% over its estimate, or whose work has got well ahead of its billing, it emails you then, instead of waiting for the weekly brief. The Money Owed page puts the cash side in one place: work done but not billed yet, open jobs whose costs have gone well past the estimate (possible change orders nobody billed), and unpaid invoices by how long they’ve been waiting.

Each finding carries a dollar impact, a severity, and a confidence level. Confidence is based on how much data supports the finding, a pattern across six completed jobs is treated differently from a sample of one, and rules that need comparison data simply don’t fire until there’s enough of it.

There’s also a Data Health page that tells you exactly what is missing before you trust a number: each job cost (materials, subcontractors, cost of goods) from the last 12 months that isn’t tagged to any job, listed biggest first, jobs with revenue but no costs recorded, time entries with no cost rate in QuickBooks, and open jobs with no cost estimate. Those gaps change what the numbers can tell you, so they’re shown rather than papered over.

The Profit Opportunity Feed compares every job you finished in the last 12 months with the ones like it, by job type, customer, job size and part of the job, and lists what to change, biggest first. Each opportunity says what the numbers show, why, what to do, how sure it is, and which jobs it comes from.

The dollar figure is worked out in the open: the price that would have hit your target margin on the same costs, minus what the jobs actually sold for. When your estimates split the price, it goes further and names the thin part. On kitchens, for example, it’s the labor: it cost nearly everything customers paid for it. Then it says how much to add to the labor on the next one.

A month-by-month view of margin by job type shows whether a change is coming through in the jobs that finish, and catches a type of work that’s starting to slip.

Every number is calculated by the application from your data. AI writes advisor notes around the opportunities, and never produces or changes a figure. Profit decisions shouldn’t rest on a number a language model came up with.

Every pending estimate in QuickBooks is checked against what your finished jobs of the same type actually cost for each dollar you charged, over the last two years. That tells you whether your pricing for that type of job reaches your target, and if it doesn’t, by how much and the price that would. When your estimates list labor, materials and subs on separate lines, each line is also set against how the cost of those jobs usually splits, so you can see which line is thin. You also see whether QuickBooks has emailed the estimate yet.

When the estimate’s lines carry quantities, it goes further and costs the job itself: hours on labor lines at your average labor cost per hour from your time entries (plus your labor burden), and products and services at the purchase cost set on them in QuickBooks. Then the check follows this job’s own price and scope, not just your pricing for the type.

The past-jobs check needs at least three finished jobs of the type. It only reads QuickBooks: you change the estimate there, and the check updates on the next sync.

When you act on a pricing opportunity, press “I’m making this change.” JobProfitAI records the margin you started from, then compares the jobs you create in QuickBooks from that day on, as they finish. You see the margin before, the margin since, and the extra gross profit, on your own jobs.

Once a week, on the day and hour you choose in your own timezone, JobProfitAI syncs the latest data and emails a short brief. It leads with the money: new margin risk on your open jobs since the last brief, estimates priced below target, and your biggest opportunities; when there’s $500 or more of new risk, or an estimate to fix, the subject line says so. Then what changed on each job: new costs and billing, margins that moved, jobs that went over their estimate. Those parts are calculated, not written by AI. A short written summary follows.

It goes to up to 10 people: you, your project manager, your bookkeeper. Each can unsubscribe with one click. And if your data is too incomplete that week to say anything reliable about profitability, the email says exactly that and lists what’s missing, instead of writing a confident-sounding summary around gaps.