RoofOS Platform

AI that answers every call and quotes the roof in 5 minutes.

RoofOS is the AI platform NeuralBuilt built for the roofing industry. Morgan answers every inbound call, takes the property address, and builds a complete line-item estimate live on the first call using real satellite roof measurements and current distributor pricing. No drive-out. No 3-day wait. No lost jobs to a faster competitor.

See how the estimate works Book a call

RoofOS is in production use by RoofOS LLC. A NeuralBuilt build.

About 5 min
from first call to a complete line-item roof estimate
2 AI agents
Riley outbound and Morgan inbound, around the clock
3-tier pricing
Standard, Quality, and Premium on every single job
Every call
recorded, transcribed, and AI-critiqued automatically

The problem

Missed calls are lost revenue. A slow estimate loses the job to whoever sent theirs first.

Two structural problems every roofing company lives with. RoofOS fixes both at once.

Inbound calls

A missed call is a missed job

When a homeowner calls and reaches voicemail, they do not leave a message and wait. They call the next company in the search results. There is no callback that recovers that moment. The loss is immediate. RoofOS answers every call, every time, with Morgan on the other end ready to quote the job.

The estimate cycle

A 1 to 3 day estimate cycle hands the job to your competitor

The traditional flow: schedule a site visit, drive out, walk the roof, measure, price, build a proposal, send it. That takes 1 to 3 days. By then, a faster competitor has already sent their estimate and the homeowner is comparing signatures. Morgan collapses that cycle to about 5 minutes on the first call.

The flagship

A complete, accurate estimate. Built live. On the first call.

Morgan answers an inbound call, takes the property address, and builds a full line-item roof estimate in real time. The measurement comes from real satellite data. The pricing comes from a state-specific material database built on current distributor rates. The homeowner hears all three tiers before they hang up.

Caller gives the address
Morgan asks for the property address. The homeowner gives it. That is the only input needed to start the estimate.
10 seconds
Satellite roof measurement
EagleView and Google Solar pull real aerial data: total square footage, pitch, facets, and roof geometry. Measured from imagery, not guessed.
About 45 seconds
Current material pricing
A state-specific pricing database returns current material costs: shingles, decking, underlayment, and accessories, built on distributor rates from SRS Distribution and QXO.
About 30 seconds
Complete estimate delivered
Three tiers, full line-item breakdown. Saved and auditable. The homeowner hears all three numbers before the call ends.
Total: about 5 minutes
Standard
Entry-level materials, full installation
3-tab or economy architectural shingles, standard workmanship warranty
Quality
Mid-grade materials, enhanced performance
Dimensional architectural shingles, 30-year manufacturer warranty
Premium
Top-tier materials, maximum lifespan
Designer shingles or metal accents, lifetime transferable warranty

Every estimate is built from real satellite measurement and real distributor-based pricing. The output is accurate enough to quote on the call, not a rough ballpark. Each estimate is saved and fully auditable in Supabase.

AI voice agents

Two agents. Every call handled.

Riley and Morgan run the phones around the clock. The command center gives you full visibility into every conversation, every score, every coaching note.

Riley / Outbound

Works the leads your team misses

Riley handles outbound calling: missed inbound calls that were not answered, dead and aged leads that went cold, and scheduled follow-up sequences. She calls, qualifies, and books appointments without anyone on your team picking up the phone.

  • Calls missed inbound leads automatically so no opportunity goes cold
  • Re-engages aged and dead leads from the pipeline with a live conversation
  • Runs follow-up cadences without manual team intervention
  • Every conversation recorded, transcribed, and logged to the CRM
Morgan / Inbound

Answers every call and quotes the job on the spot

Morgan is the AI estimating assistant that never misses an inbound call. She takes the property address, pulls real satellite measurements, fetches current distributor pricing, and delivers a complete three-tier estimate in about 5 minutes on the first call.

  • Answers inbound calls around the clock, no voicemail, no missed opportunities
  • Pulls real roof measurements via EagleView and Google Solar APIs
  • Prices from a state-specific material database built on SRS Distribution and QXO rates
  • Delivers Standard, Quality, and Premium estimates live on the call

Every call recorded, transcribed, and AI-critiqued

The command center captures every conversation, transcribes it, and runs an AI critique: what the agent handled well, where the conversation could have been stronger, and how the call compares across your team. No call goes unreviewed. Your operation gets coaching without coaching sessions.

"Had you not told me, I would have thought it was a real person."

A tester, after a live call with Riley

The CRM and estimator

iPad-first estimates. 3 tiers. Your pricing, your margins.

The roofos-app is the multi-tenant SaaS side of RoofOS: a CRM and estimator built for roofing companies that want to own their pipeline, their pricing logic, and their customer data.

RoofOS estimator interface showing a roofing job with three-tier pricing: Standard, Quality, and Premium, with full line-item breakdown

What the CRM delivers

  • Three-tier estimate creation on every job: Standard, Quality, and Premium, each with a full line-item breakdown
  • Per-tenant material library so each company's shingle, decking, underlayment, and accessory pricing stays separate and configurable
  • Labor rates and markup percentages set per tier and per company, not locked to a platform default
  • Lead pipeline with stages from new lead to signed job, with follow-up tracking and next-action reminders built in
  • iPad-first interface designed for field use, not just the office
  • Every estimate auditable: created by Morgan on the call, or built by a rep in the app, always saved with a full trail

calc.roofos.co

How much is your voicemail costing you?

Before a roofing company looks at RoofOS, they see their number. The loss calculator shows exactly how much unanswered calls cost in annual revenue, based on four inputs any owner already knows. It opens the conversation with math, not a pitch.

Average job ticket$12,000
Revenue per completed roof job
Inbound calls per week25
Total calls your number receives
Percent unanswered30%
Calls that go to voicemail or ring out
Close rate25%
Percentage of leads that become signed jobs
Annual revenue leakage
$1,170,000

About 98 jobs lost per year from missed calls alone, based on your inputs

Run the full calculator

The engineering

AI grounded in real data. Built end to end.

RoofOS is strong because the AI cannot hallucinate the measurement or the price. Every estimate starts with a real data source, not a model making its best guess.

Application layer

Next.js and Supabase

The ops platform at ops.roofos.co and the multi-tenant CRM both run on Next.js with a Supabase backend. Real-time data, row-level security, a fully auditable estimate history, and no vendor lock-in on the data layer.

Telephony

Twilio

Every call, inbound and outbound, runs through Twilio: routing, recording, and the streaming audio connection to Morgan and Riley. Full call logs, every session. No voicemail slip-through.

Voice synthesis

ElevenLabs

Morgan and Riley speak using ElevenLabs voice synthesis. Realistic, low-latency, consistent across every call. Homeowners have a full conversation, not a phone tree.

Speech recognition

Deepgram

Deepgram transcribes every call in real time. The transcript is the source for AI critique, CRM logging, and the auditable call record. Fast and accurate enough for phone audio quality.

Reasoning

Claude (Anthropic)

Claude is the reasoning engine: it parses the conversation, decides when to trigger measurement and pricing lookups, and assembles the final estimate. It is grounded by real data at every step so the output is accurate, not a confident-sounding guess.

Measurement

EagleView and Google Solar

Real satellite and aerial roof measurement. EagleView delivers total square footage, pitch, facets, and geometry from real imagery. Google Solar supplements with roof plane data. Both are verified measurement sources, not approximations.

Distributor pricing

QXO and SRS Distribution

Material costs come from a state-specific pricing database built on real distributor rates from SRS Distribution and QXO, kept current rather than guessed. This is what makes the estimate accurate enough to quote live on the call.

Prospect tool

calc.roofos.co

A standalone calculator at calc.roofos.co. Four inputs, one output: the annual revenue leaking from unanswered calls. Opens the RoofOS conversation with a number the prospect already cares about, before they have seen the platform.

Built end to end

Built in house, end to end

RoofOS was designed, engineered, and shipped end to end: no agency, no subcontractors, no off-the-shelf SaaS kit under the hood. Every component was purpose-built for roofing, and the platform owns its data and its infrastructure.

The reason RoofOS estimates are strong is the engineering logic: the AI cannot hallucinate the measurement or the price. EagleView and Google Solar give Morgan the real roof geometry. SRS Distribution and QXO pricing gives her the real material cost. Claude assembles those verified inputs into a line-item estimate. The output is accurate because the inputs are accurate, not because the model guessed well. That is not how most AI software is built. It is the difference between a tool a roofing company can quote from live on the first call, and a tool they have to sanity-check before they send anything.

How we got here

We did not show up one day. We put in the work.

RoofOS did not start in a strategy deck. It started on a roof, on the phones, and inside a real roofing company.

It began with Colorado Front Range Roofing in Pueblo, owners Jim and Shawn, where the first estimator was built for a real crew with real jobs on the line. To understand the work at ground level, Trey spent more than twelve hours on an actual roofing job site with Diego, a working roofer: boots on, start to finish, not research from a laptop. Then came the phones. More than 150 roofing companies called in a single week to learn exactly where they were losing money. Luke, who cold-calls for a living, ran a market test posing as a homeowner across 60-plus Florida roofers. Two of them followed up within ten days. That silent leak is exactly what RoofOS was built to close.

Field research

150+

Roofing companies called in one week to find exactly where leads were being lost.

Market test

2 of 60+

Florida roofers that followed up within ten days when Luke posed as a homeowner. The rest went silent.

On the roof

12+ hrs

A full day on a real job site with a working roofer, building for the job as it actually is.

Riley, judged by a pro

~2.4s

Median voice response. Luke, who hears hundreds of calls a year: "Even Riley V1 is better than 80% of calls I've been on."

Custom software for your business

See what custom software like this could do for your business.

RoofOS is what NeuralBuilt builds: AI systems grounded in real data sources, not bolted onto a generic platform. If speed, accuracy, and complete ownership of your data are what your operation needs, send a note.

trey@neuralbuilt.ai