← Projects

QBO-P02 · Quobo Labs fleet

Morgan Land Co.

Labs partnershipPre-launch

A Lafayette landscape company the lab invested in with code: a homeowner gets an AI-measured price from their own lot in about 90 seconds, while a designer studio, a constraint-solving scheduler, and a customer portal run the rest of the business on one data model.

Feature-complete and counting down to launch — the story is below.

Chapter one

The problem

A residential landscape company grows or stalls at the quote. Industry research says eighty percent of customers hire whichever company quotes first — and the local standard is a phone call, a drive-by site visit, and a number in one to three days. Every hour of that lag is a job leaking to whoever answered faster. Worse, the person doing the measuring is usually the owner or the best designer — the most expensive people in the company, spending their afternoons pacing lawns with a measuring wheel.

Behind the quote sits the second throttle: operations run on memory. Which crew carries the state-licensed pesticide applicator, whether rain will wreck tomorrow's fertilization run, which jobs cluster on the same side of town, who never got their day-before text — for most local operators that lives in the owner's head, a whiteboard, and a few disconnected apps that don't agree with each other. The tools on the market record work after a human decides it; none of them lets work book itself around license, weather, equipment, and drive-time constraints.

Quobo Labs' answer was not to sell this company software. It was to invest in the company by building the software — the lab's capital is code, and the thesis is a local operator with a structurally unfair speed and margin advantage.

The mechanics

How it works

  1. A homeowner quotes their own lawn

    One screen captures a name, an email, an address — and the moment the address lands, the measurement pipeline starts. A parcel polygon comes from the parish assessor's public records, aerial imagery is fetched, and a vision model returns the outlines of lawn, beds, hardscape, and driveway — outlines only, never areas; the app does the geometry itself and rejects any result that claims more surface than the lot holds. The customer watches polygons draw over their own roofline, and if the frame misses, a crop tool lets them drag a box around their lot — the live square footage recomputes with the same math, though a customer's box can inform an estimate, never bind a price.

    The demo pipeline mid-quote — six measured segments, each wearing its own confidence grade
    The demo pipeline mid-quote — six measured segments, each wearing its own confidence grade

  2. A price that knows its own confidence

    A deterministic pricing engine — no randomness, every cent traceable to a cost-up methodology — prices the scoped work from a seeded catalog with modifiers for slope, access, flood zone, and canopy. Confidence is the weakest measured segment, never an average: clear the bar and the quote is binding within four percent; land in the middle band and it's a stated range; fall below — or trip any of thirteen routing rules that watch for design-build language, estate-scale work, drainage, and licensed trades — and the system books a conversation with a human instead of pretending it can quote. Every routing decision records which rule fired.

    A binding quote from the demo catalog — SKU-coded line items, the stated range, a price the business honors
    A binding quote from the demo catalog — SKU-coded line items, the stated range, a price the business honors

  3. Booking closes the loop

    An eligible quote flows straight into a slot picker fed by the real scheduling engine, and a deposit secures the visit. An abandoned quote quietly earns a signed thirty-day resume link; a missed phone call to the business triggers a text with the quote waiting at the other end. The lead never depends on someone remembering to call back.

  4. Complex work gets a designer, fully briefed

    An escalated lead opens in the studio with everything the pipeline already learned: the parcel, the measurements and their confidence, the flood zone, the customer's own words, and the reason it was routed. The designer draws zones on a canvas over the property's real aerial tile, prices in phases from the same catalog, generates concept renderings seeded with the property's own imagery, and publishes a private proposal microsite the client reads at the kitchen table — while engagement tracking shows which phase they lingered on.

    The studio's site canvas: zones drawn over a property's real aerial imagery

  5. The workbench runs the operation

    The admin surface is the company's operating system: a lead inbox with AI qualification scores and routing reasons, a drag-and-drop pipeline board split between maintenance and design-build, and a scheduling optimizer that treats licensing, equipment maintenance windows, temperature, and rain rules as hard constraints — then scores what remains by crew continuity and drive time, offering a placement and two alternatives, deterministically. Dispatch prints per-crew day sheets, arrival texts fire as trucks approach, and a job cannot be marked complete without before-and-after photos — a rule enforced in the database, not in a training binder.

  6. The customer keeps a key to their own account

    A homeowner signs in and manages the relationship themselves: pause or change a plan, add services, reschedule a visit against the crew's real availability, read invoices, leave a review. A referral link attributes a neighbor's quote and pays a credit into a real ledger the customer can spend on their next invoice. Underneath all five surfaces, every table is access-controlled, every change lands in an immutable audit log, and every AI call is recorded with its confidence and its cost.

Field notes

Would this fit your life?

A homeowner with three competitors' tabs open

The moment
She wants weekly mowing and a mulch refresh, and clicks "Get an instant quote" expecting a contact form.
What they do
One screen takes her address; polygons draw over her own lawn while she picks services from guided cards. Every measured segment clears the confidence bar, so the price comes back binding — book it and it's honored — and she takes a Saturday slot with a deposit.
The payoff
A real, honorable price in minutes, while the competition is still scheduling site visits.

An estate owner who wants a master plan

The moment
Outdoor kitchen, lighting, drainage — his free text and his neighborhood trip the routing rules, and no robo-price is ever shown.
What they do
A designer opens his file with the property already measured and profiled, drafts three phases on a canvas over his actual aerial tile, and sends a private proposal site with concept renderings of his own backyard.
The payoff
The high-margin lead never got an insultingly automated number — and the designer never re-typed a thing.

The dispatcher at seven a.m.

The moment
Two self-bookings landed overnight, and rain moved into Thursday's forecast.
What they do
The optimizer already placed both bookings on crews that satisfy every license and equipment rule; the exceptions view flags Thursday's fertilization visit against the rain rule, and she drags it to Friday. Day sheets print; arrival texts handle themselves.
The payoff
A week of jobs placed without a single licensing, weather, or capacity violation — by construction, not by vigilance.

The owner, between site visits

The moment
He wants to know what came in and what's stuck.
What they do
The inbox streams scored leads, each wearing the reason it was routed. He triages in minutes, watches win rate by neighborhood on the pipeline board, and can pull the audit trail on any change anyone — or any AI — ever made.
The payoff
The voicemail box became a scored, reasoned queue; "how's the pipeline?" became a screen, not a feeling.

No varnish

From the lab notebook

Where it stands

Feature-complete and pre-launch. All five surfaces — brand site, self-quote, studio, workbench, portal — run end to end with no placeholder pages, deployed as a working demo. Behind them: roughly two thousand tests that pass with zero API keys, live-database suites including a cross-customer isolation gate, browser-automation specs, and a CI pipeline that hard-fails if the database suites can't run. What hasn't happened yet is the business itself: no live customers, no field-verified measurements, no revenue through the system. The machine is built, tested, and waiting.

Waiting on keys, not code

Every paid vendor — imagery, payments, texting, email — sits behind a stub with the same interface as the real thing, and the stubs flow through production code paths: the demo's fake checkout completes by posting a signed synthetic event through the real payment webhook. That's why the whole system demos green with no keys, and why go-live is a matter of opening vendor accounts rather than finishing features. It also means the marketing site still says "Photo coming soon" where real project photography belongs.

What's still rough

The hardening program went deep on the core demo spine but has covered a minority of the full feature list so far, leaving shared libraries less scrutinized than the flows that use them. Proposal acceptance is click-to-accept, not yet a true e-signature. The admin's deep links all work, but there's no dashboard home page tying them together. The docs keep an unvarnished ledger of this — including the bugs the hardening loop caught, like a booking check that quietly rejected every eligible quote until it was found.

The investment thesis

The lab chose the expensive path on purpose: own operations end to end rather than sync to rented tools, and build the measurement capability in-house — about two cents per measurement, with the geometry computed in the app — instead of paying a vendor per lookup. Nobody builds a constraint-solving scheduler and a payments spine for a services fee; you build them when the return rides the business's own results. And the long game is a dataset: once live, every completed job feeds field-verified measurements back against the AI's estimates, accumulating a one-parish training set no national vendor can replicate.

Next in the fleet

QBO-F01

Quobo Industrial

In production

Your facility, finally answerable: Quobo's flagship platform models the plant itself — structure, dependencies, live readings, operator knowledge — so a grounded AI can propose work to a named human who always makes the call.

Continue the tour →