Wasson Landscape & Nursery · est. 1977

An AI workforce,
built in-house,
one agent per role.

Our agents read Aspire, the drawings, and the shop floor. They measure, check, flag, and learn. People make every call that touches money, records, or a client.

0
estimating rules learned from our own jobs, each one approved by Dan
0
live Aspire catalog items the brain prices from, synced every night
0
kit recipes expanded before any omission check runs
0
design plans read, measured, and reviewed by the engine so far
The family

Seven agents. Each one knows its job, its person, and its limits.

Every agent is read-only against our systems. Sign-in is required for anything that shows a number.

How the estimating advisor works

A drawing and an estimate go in. A verdict comes out.

1

It reads the design

Layer by layer, straight from the Land FX plot. Paver fields, concrete, coping, steps, pools, beds, lawn, every plant symbol, every light. It cross-checks the scale bar and looks at the rendered sheet with its own eyes.

2

It reads the estimate

Pulls the Aspire estimate by number, rolls up change orders, expands every kit into its recipe, and reads the estimator's notes before it judges anything.

3

It advises, never decides

Margin panel first. Then Blocking, Flagged, and Passed findings, what is missing on the drawing, and what is good. The designer decides what to change. Nothing is ever pushed to Aspire.

The learning loop

Every submission teaches it. Every lesson waits for a human.

Behind each report, the engine quietly prices the same design on its own and measures how far it landed from the team. The gap becomes lessons. Lessons become proposed rules. Dan approves them, one by one.

  • Rule
    Never a $0 line for custom stone

    Vals Stone is priced differently on every job, so the engine now carries a labeled allowance from 45 pooled estimates and always asks for the real quote.

  • Rule
    Size the concrete allowance

    While the sub question is open, the engine prices net square footage at the team's own standard and steps at 2.5×, instead of a token number.

  • Learned
    Rates come from our history

    Porta potty rates, Vals shipping, install hours per hundred square feet: pooled nightly from every Aspire estimate, never typed into a rule.

  • Heads-up
    Flags, not blocks

    When something could vary, like screen-tree sizing, the engine states the standard after the scan and lets the designer decide. Training, not gatekeeping.

Non-negotiables

Built like the asset it is.

Reads, never writes

Every agent is read-only against Aspire, the catalog, and the books. Nothing reaches a client or a record without a person signing off.

Asks, never guesses

Missing scale, unlabeled areas, concrete ownership, new versus existing plants: the engine asks a targeted question. A confident wrong number is the one unforgivable failure.

Learns, with approval

Corrections become rules with provenance: who taught it, which job, which day. Dan approves every rule before it touches an estimate.