AI-Enabled Proposal Builder, From Roof Inspection to Proposal

The CRM fixed operations. Sales still ran through a disconnected tool.

Once the Roofing CRM shipped, a job had one home from the moment it was won to the moment it was paid. But a job still had to be won first, and that entire front half, the assessment and the proposal, happened outside the CRM entirely, inside a third-party platform called Roofr. Roofr did its one job well: it measured roofs and produced proposals. But it had no relationship to the CRM. A sales rep who'd just qualified a prospect inside the CRM had to leave it, rebuild the same customer and job in Roofr, run the assessment there, and, once a homeowner approved, manually carry the outcome back. The CRM had solved the job's life in production. Its life before that still lived somewhere else.

A second system, disconnected the same way the first ones were

This was the same seam problem the CRM had already solved once, reappearing on the other side of the job. Four fault lines showed up.

We audited Roofr's workflow, then traced one proposal end to end.

Before designing a replacement for Roofr, I traced a real proposal end-to-end, from the moment a rep booked an assessment to the moment a homeowner signed, and sat with the roles on either side of it. The goal wasn't a feature-parity list against Roofr; it was seeing exactly where the handoff to the CRM broke down.

Build the front half into the CRM, not bolt it on

The first real decision wasn't a screen. It was whether to integrate with Roofr's API or absorb assessment and proposal work into the CRM's own data model. An integration would have kept two systems in sync forever; it wouldn't have removed the seam, just automated the stitching. We chose to build a native module that shares the CRM's job record directly, so a proposal isn't synced to a job, it is the job, at an earlier stage.

One continuous path: inspection to signature

A prospect leaves the CRM's qualifying stage as a scheduled assessment and doesn't surface again as a "won job" from somewhere else. It moves forward as the same record, through one flow, until a signed contract hands it to production.

Six modules, two groups, no separate app to switch to.

The navigation splits into the daily selling work, the configuration behind it, and the admin layer that closes the loop, each one reading and writing the same job the CRM already owns.

Gives reps visibility into their pipeline, assessments completed, proposals sent, proposals approved, and projects closed, plus recent activity at a glance.

Inspect properties, record damage items, capture and organize photos, and upload supporting documents in one structured workflow.

A flexible builder for generating customer proposals, customizing sections per roofing service, and keeping every field linked to the CRM job it belongs to. A rep chooses upfront whether the proposal is linked to an assessment or standalone, and an inline AI agent can draft the entire proposal from a prompt either way, then refine individual sections on request as the rep edits.

Templates keep a proposal within a specific category so it can be created faster. A rep can use one as-is, modify it, or create a new one from scratch.

The shared catalogue behind both estimate types: itemized, a line-by-line breakdown of items and price, and roof system, a build-specific estimate. Every catalogue entry carries its own set of instructions and price.

Centralized configuration for how assessments and proposals behave across the team.

Once a proposal is approved, rep and homeowner complete the agreement digitally, and it becomes part of the connected project inside the CRM.

The slowest part of a rep's day was never the pricing or the signature. It was staring at a blank proposal after a long inspection. A rep can now describe the job in a prompt and get a complete draft back, whether or not that proposal is linked to an assessment. The agent pulls from the assessment when one exists, and works from the prompt alone when it doesn't. From there, the agent works inline: a rep can ask it to rewrite a single section, adjust scope language, or re-price a line item, without starting the proposal over.

Rewriting a paragraph was never the whole job. A proposal often needs a section added for a service the homeowner asked about, one removed because it doesn't apply, or the order changed to match how the rep is presenting. The always-on AI editor handles this the same way it handles wording: on request, in place, without the rep leaving the builder.

Shipped and still evolving, here's where it stands today

This has already launched and is in active use by the sales team. It isn't a pilot. From here, the work is ongoing: testing with real reps, redesigning what usage shows us doesn't work, and improving the experience iteratively. So the honest version of "impact" here is a status report: what's shipped, what's being refined right now, and what's still ahead.