Wave
Docs menutap to expand

Concepts

Wave Intelligence

Wave reasons about the job: property type, water category, scope. The estimate reflects what an experienced estimator would build.

Last updated

Anyone can build a chat front-end that adds a drywall line item when you type "I tore out drywall." That's a string match. It's not what Wave is doing.

Wave reasons about the job.

When you describe a loss, Wave infers context: what kind of property this is, the water category, how the structure is built, what an experienced estimator would put on a defensible bid. The estimate that comes out the other end isn't a transcription of what you said. It's what someone who's run a thousand jobs would write down for this job.

That's Wave Intelligence. The rest of this page is what it looks like in practice.

It knows what comes with the work

Tell Wave a Cat 2 supply line failed and soaked the hall bathroom and hallway, and the estimate includes the work that comes with that loss in a home: the tear-out that has to happen, the cleanup the standard requires, the protective equipment for the crew, the containment because the loss crossed a doorway. You didn't list those. Wave knew they belong, because it knows what a residential structure is made of and what a Cat 2 loss demands.

Category-aware adjustments

Cat 1, Cat 2, and Cat 3 are not labels Wave attaches and forgets. They reshape the estimate. Protective equipment, cleaning and application steps, containment, and documentation all scale with the contamination level, so a Cat 3 estimate looks materially different from a Cat 1 estimate for the same rooms. Change the category and Wave updates its view of the job, not just a label.

Scope-aware totals

When you add a room mid-estimate, per-room totals on existing rooms can shift slightly. People sometimes ask whether this is a bug. It isn't.

Some items are sized against the whole project, not any single room. Equipment is the obvious one: a machine serving three rooms isn't three separate line items, it's one item whose cost allocation changes as the room count changes. When scope changes, Wave re-evaluates the project as a whole, the grand total moves up because you added scope, and the per-room math redistributes. An experienced estimator would do the same math in their head.

Defensive bidding

One observable consequence of Wave Intelligence is that estimates are defensively bid: Wave errs in the direction of including more line items, not fewer. Forgetting one line item can wipe out the margin on a job; including one you didn't need costs a few seconds to delete. The fuller explanation lives at How Wave thinks about scope.

Why this is hard to replicate

A chat-to-form-filler is a string-matching exercise. You can build one in a weekend with any LLM.

Wave isn't that. The reasoning is grounded in restoration knowledge: industry drying standards, structural anatomy, regional pricing reality, the specific way carriers actually approve estimates. It's the difference between a chatbot that types "drywall repair" into a form and a system that knows what comes with drywall repair at this category and this scale.

That's why Wave produces estimates an experienced estimator would have produced, not estimates a model trained on internet text would have produced.

Related