$200k and up
What one competent engineer costs you for a year, fully loaded. One engineer. Not a team, and not somebody who already understands your trade.
Technology Partner
They're not better at the work. They're better at everything around it.
You already know the national outfit across town does not run a better crew. What it has that you do not is an engineering department: people whose entire job is making the business work better next quarter than it does now. For fifty years that was a headcount problem. It is not anymore, and what deploys into your business now is not one engineer. It is the whole department.
Start Here
Imagine Amazon opened a heating and cooling company tomorrow. Not a marketplace for contractors. An actual company, with trucks and technicians.
It would not send better technicians. The condenser goes in the same way. What it would bring is the part nobody sees from the driveway: dispatch that routes itself, pricing that updates itself, a customer record that follows the job, and an engineering organization behind all of it whose only purpose is to make next quarter's version of the business run better than this one's.
That is the whole advantage. Not the trade. The engineering. And it was never available to you at any price, because the smallest useful amount of it cost more than the business clears.
You were never losing to a better contractor. You were losing to an engineering department.
Why you could never hire one
$2k to $8k a month
What a business your size already pays for field software, booking, payments, a marketing retainer, a bookkeeper, and the glue between them. None of it compounds. Next year it does the same things it does today.
A login
So you rent average software, priced by splitting one engineering team across ten thousand businesses that are nothing like each other.
None of that was anyone being unfair to small businesses. It was arithmetic. The arithmetic is what changed.
What Changed
The capital budget for technology is now token cost
The bill for building software used to be salaries. Now it is mostly compute. That is the whole change, and it is worth sitting with, because the salary line is exactly what put an engineering department out of reach. Not the code. The headcount.
That line item moved. The cost of writing and maintaining software is collapsing toward the cost of the compute that produces it. A budget that used to be a salary is now a bill you can read in one line, small enough for a three-truck shop to carry.
When the input to an engineering department stops being headcount, the size of your business stops deciding whether you get one.
Who Deploys
Not an engineer. A department.
Enterprise software has a job title most people outside it never hear: the forward-deployed engineer. Instead of shipping a product and a manual, the vendor puts an engineer inside the customer's operation to watch how the work actually happens and reshape the software around it. It is the most effective thing anyone has invented in software, and the most expensive, which is why it has always stopped at the enterprise line.
Senior judgment at the top. Agents underneath.
What deploys into your business is not one hired engineer. It is the shape of a department. People work at the altitude a large company reserves for its most senior engineers: deciding what the business should do, what good looks like, and where the line is. The building and the daily operational work happen beneath that, continuously, without a salary attached to each one.
The engineer
Builds and runs your site, your booking, your payments, and the wiring between them. When you need something changed, it changes that week, not next quarter.
The front desk
Answers the text, books the slot, sends the confirmation, and chases the no-show. It does not go home at five.
The marketer
Writes the offer, sends the campaign, and follows up with the customers who quietly stopped coming back.
The collections clerk
Works the failed card, the expired membership, and the invoice nobody wants to call about. You keep the relationship. It handles the awkward part.
The analyst
Knows what you are actually making, which customers are about to leave, and which service only looks profitable.
The migration crew
Gets your customers, your history, and your calendar out of the old system and into this one. Once, properly, without you retyping anything.
Every role runs a loop against a named outcome
Agents left to wander are a demo. Agents pointed at one narrow, measurable outcome are staff. Each role on your deployment runs a loop, and every loop has an outcome somebody would notice if it stopped.
01
Fill tomorrow.
Find the open slots in the next 48 hours, find the customers who are due, offer them the time, book it, confirm it.
02
Never miss the call.
Pick up the one that comes in after hours or while every line is busy, work out whether it is an emergency, and get it on the board before the caller tries the next name on the list.
03
Collect what is owed.
Find the failed card, the expired membership, and the invoice past due. Work each one until it clears or until it genuinely needs you.
04
Bring back the quiet ones.
Find the customers who used to come and stopped. Make each one a specific offer, not a newsletter.
05
Close the week.
Reconcile what was booked, what was worked, and what was paid, then tell you the three things that did not match.
Every loop is logged and every loop is measurable, and you decide which ones are allowed to run without asking you first.
The Reimagining
What a powerhouse would do with six trucks
Not better technicians. The same crew, the same vans, the same territory. What changes is what the business is optimizing for, and a company with a real engineering organization optimizes for things a six-truck shop has never had the tools to see.
Density is the whole game.
A shop dispatches by who is free. A logistics company dispatches by who is close, then goes further and shapes demand to create the closeness, offering Thursday to the customers already sitting near Thursday's route. Four jobs a truck a day and four and a half jobs a truck a day are the same crew and a different company.
Demand is forecastable, so stop reacting to it.
Heating and cooling demand is set by weather, and weather is known a week out. The first cold night produces a week of no-heat calls that everybody treats as a surprise. It is not a surprise. It is a schedule you could have filled in September and a truck you could have stocked in advance.
The maintenance agreement is the actual product.
Every large consumer business eventually discovers the same thing: the subscription is not a revenue line, it is the reason the customer comes back to you instead of searching. Most shops keep agreements in a spreadsheet and treat them as paperwork. They are the thing that smooths a seasonal business and puts you in the basement on the day the equipment finally dies.
A missed call is not a missed job.
A furnace lasts fifteen to twenty years. The call that rolled to voicemail at nine at night was not a lost service ticket. It was a two-decade relationship handed to whoever answered, plus every replacement, referral, and tune-up inside it.
Price the customer, not the job.
A tune-up looks like a low-margin visit if you price the visit. It is the cheapest way ever invented to be standing in the basement, with the equipment history in hand, in the year the unit fails. Shops price jobs because jobs are what they can see. An engineering organization can see the customer.
None of that requires a better technician. It requires somebody whose whole job is looking at the business this way.
Use Case 01
Acme HVAC, with the arithmetic shown
Acme HVAC is a composite, not a customer, and what follows is a model rather than a measured result. The arithmetic, the assumptions behind it, and the things it deliberately leaves out are all below, so you can run it against your own numbers and find the places it does not hold. We would rather hand you a model you can check than a number you have to believe.
The Shop
- Trucks
- 6
- People
- 11, counting the office
- Work
- Residential service and light commercial, plus a 24/7 emergency line
- Maintenance agreements
- About 400 on the books
- Jobs per truck per day
- 4.0
- Working days a year
- 250
- Average ticket
- $340
- Revenue
- About $2.1M, counting agreement billings
Second generation, founded in the early seventies. The owner's brother runs installs. His nephew does dispatch, when he is not on a roof.
Where The Money Leaks
- The emergency line rolls to whoever is on call, and half the time they do not hear it.
- Dispatch is by who is free, so two trucks cross the same town most mornings.
- Agreements renew when somebody remembers to open the spreadsheet.
- Failed cards on monthly agreements turn up in the bank statement a month later.
- The owner quotes jobs at nine at night, because that is when the phone stops.
The Arithmetic
Fill tomorrow.
Route density: 4.0 to 4.6 jobs per truck per day.
0.6 × 6 trucks × 250 days × $340
$306,000
Never miss the call.
Roughly 8 after-hours and overflow calls a week, 40% of which would have booked.
8 × 0.4 × 52 weeks × $340
$56,600
Collect what is owed.
About 8% of 400 agreements fail on a card each year, 60% recovered, each worth its billing plus the service it pulls through.
400 × 0.08 × 0.6 × $840
$16,100
Bring back the quiet ones.
About 500 past customers gone quiet two seasons or more, 6% reactivated.
500 × 0.06 × $340
$10,200
Less, because two of those overlap
Recovered calls have to land somewhere, and the only slots we created are the ones in the first line. Counting both in full would bill the same truck-hour twice.
166 recovered jobs × $340, already counted above
($56,600)
About $332,000 a year, on a $2.1M shop. Call it 15%, without a seventh truck and without a twelfth person.
The uncomfortable part of that model is where the money is. Route density is worth thirty times the reactivation campaign, and reactivation campaigns are what gets sold to shops like this one.
What this model leaves out
- This is gross revenue, not profit. Parts, fuel, and any overtime on the extra jobs are not netted out.
- It assumes the demand is already there and being turned away. If your calls are all being served today, tighter routing lowers your cost to serve, it does not raise your revenue. Ask us to run it against your own turned-away count.
- It is steady state. Nobody hits these numbers in month one, and a seasonal trade does not hit them evenly across the year.
- Every input is a single number, not a range. Halve the density gain and the total falls by roughly $125,000.
Assumptions
- Average ticket of $340 and 250 working days are typical for residential service HVAC. Yours will differ.
- The density gain assumes clustering plus demand shaping, which is the ordinary result of scheduling against a route rather than against a calendar.
- The 40% booking rate on recovered calls is deliberately conservative for after-hours no-heat demand.
- Agreement value of $840 counts the billing plus the service work an agreement holder pulls through in a year.
- No price increase is assumed anywhere in this model.
Change any number here to your own and the model still runs. That is the point of showing it.
Narrow On Purpose
Deployed into a trade, not configured into one
The reason general business software feels wrong is that it is general. It arrives knowing nothing about your trade and asks you to teach it, one settings page at a time. You become the systems integrator, and eventually the system itself is the thing in your way.
SmartFront deploys per trade. An HVAC deployment knows about maintenance agreements and seasonal load. A gym deployment knows about memberships, classes, and the member who has not scanned in for three weeks. A law firm deployment knows about intake and conflicts. The narrowness is not a limitation. It is what lets the loops mean anything.
BarberCleaningDaycareDentalElectricalGymHVACLawPlumbingRestaurantSpaTaxTherapy
A vendor sells you a tool. A department does the work.
Software you bought
A department that deployed
Software you bought
Onboarding is a checklist you complete
A department that deployed
Your business is set up before your first login
Software you bought
Something breaks, you file a ticket
A department that deployed
Something breaks, it is already being worked
Software you bought
You need a feature, you upvote it
A department that deployed
You need a feature, it gets built into your account
Software you bought
After hours it goes to voicemail
A department that deployed
After hours it is still working
Software you bought
The software records the work you did
A department that deployed
It does the work and shows you what it did
Software you bought
Leaving means exporting a CSV and good luck
A department that deployed
Your data is yours, whole, whenever you want it
Where This Goes
Four things we think happen next
01
The floor moves, and that is the actual event.
Custom software has always had a size floor. A company had to be big enough to justify engineers, and everyone underneath it got templates. AI is not mainly making software better. It is moving that floor down, and the businesses sitting just below the old line are the ones whose position changes most.
02
Renting average software stops being the only option.
Vertical software exists because one engineering team could be split across ten thousand roughly similar businesses. That split was the product. When building for one business stops being expensive, the premium for being averaged in with everyone else gets harder to defend.
03
The scarce thing stops being engineers and becomes judgment.
When building is cheap, the bottleneck moves to knowing what to build and being accountable for whether it worked. The valuable technology role inside a small business will not be a person who writes code. It will be someone who owns an outcome.
04
Software stops being something you look at.
Dashboards were a consequence of scarce labor. Software could only record what people did, so it got very good at displaying. That constraint is gone, and the question moves from what happened to what got handled.
And one prediction against ourselves.
Cheap software is going to be available to everyone, your competitors included, so building it will not stay a differentiator for long. What stays hard is the deployment: knowing one trade well enough to point the work at the right outcomes, and staying inside the business while it changes. That is the part we are betting on. It is also the part you should judge us on.
Where We Are
SmartFront is in early access. A small group of businesses are already running on it while the department gets better at deploying. If you would rather have people inside your business than another tab open in your browser, tell us about the business and we will tell you honestly whether we can deploy into it yet.
Reasonable objections
Why not hire a freelancer, or just build it myself with an AI tool?+
You should try it, and a fair amount of what we do started exactly that way. The thing worth knowing going in is that building the first version is now the easy half. The hard half is the second year, when somebody has to keep it running as an API changes, as your prices change, as the person who built it moves on. A contractor hands you a system and leaves. Then the system is yours to maintain, and eventually it is the thing standing in your way. Deployment means nobody hands it back to you.
Is this just a chatbot?+
No. A chatbot answers a question. This takes actions in your business: it books, charges, texts, refunds, and publishes. A chat window is one door into it, not the thing itself.
What if it does something I did not want?+
Every action is logged and reversible, and you set the line between what runs on its own and what waits for your approval. Most owners move that line outward over time. Nobody has to.
Do I have to change how I run things?+
That is the part deployment exists to avoid. The department learns your process. You are not supposed to learn its schema.
I am not technical.+
That is who this is for. If you were technical you would have built some of this already, and you would know exactly how much of it is left.
Something goes wrong at six on a Saturday. Who picks up?+
A person, and the same one each time, not a queue. The department runs your business, so when it does something you did not want, that is ours to unwind, not yours to file a ticket about. Ask us who yours would be before you sign anything, and if the answer is a support address, that is a real answer too.
How long until it is running?+
Days, not quarters. The slow part is usually getting your data out of whatever you are on now, and that is the migration crew's job, not yours.
What does it cost?+
Plans are on the pricing page. The honest framing: compare it against the stack of tools and retainers it replaces, not as another line added on top of them.