Home Services · Operations + AI

Radon1

Radon1 had a staff member spending up to two hours every day stitching together a manual ops report. We replaced it with a live HouseCall Pro dashboard and a Claude-powered analysis layer — and stayed on as the team's managed AI partner.

Client

Patrick Neuman · Radon1

Radon mitigation · multi-market

Engagement

Build + managed retainer

Surface

Dashboard · Daily digest · AI agents

Status

Live · reports.radon1.com

2026 — ongoing

01

The problem

Radon1 is a radon mitigation business operating across three markets with field crews running off HouseCall Pro. The daily question every operator asks is simple: "how did we do yesterday?" Revenue per technician, jobs booked, troubleshoots required, what's in the pipeline, what needs follow-up.

The answer used to require a staff member named Ben spending one to two hours every morning pulling reports out of HouseCall Pro, formatting them in Excel, and emailing the result around. By the time Patrick saw the numbers, they were already several hours stale. The information was useful — but the cost to produce it was a full-time human pulling roughly 500 hours a year of manual reporting out of the business.

On top of that, Patrick was trying to figure out where AI fit in his operation. He'd been pitched dashboards before. He'd been pitched chatbots before. He'd seen nothing that survived contact with how a multi-market home services business actually runs.

02

The thesis

Two engagements stacked together. First, a one-time build: connect to HouseCall Pro's API, replace Ben's manual report with a live web dashboard at reports.radon1.com, automate the daily email digest, and surface the metrics Patrick actually wanted to see — revenue per tech, daily averages, troubleshoots over time, team performance.

Second — and this is the part that matters — a $1,900/month managed retainer. Not "we built it, call us if it breaks." The retainer covers ongoing dashboard operations, plus an expanding set of AI capabilities scoped specifically to Radon1: a Field Tech Support Agent grounded in AARST/ARSH standards with citations, a Business Intelligence agent that can answer ad-hoc questions about the operation, and continuous improvement as Anthropic's models and tooling evolve.

The pitch we made Patrick: "The AI tools underneath your business are changing weekly. My job is to stay on top of all of it so you don't have to. You get the latest and greatest pushed into your setup without needing to follow the AI space yourself."

Claude is an assistant you talk to. This is an operator that works for you — connected to your actual systems, pushing information to you instead of waiting for you to ask.

Proposal note to Patrick, Mar 2026
03

The build

The dashboard is a Next.js application on Vercel that pulls from the HouseCall Pro API every morning, normalizes the data, and renders it as a series of operational views. Daily report fires at 6am Central into Patrick's inbox. Weekly team report rolls up performance across all three markets. Troubleshoots-over-time view tracks the quality signal that matters most in radon work.

On top of the dashboard sits an analysis layer powered by Claude. When Patrick asks "how did Nashville do last week?" or "which tech has the highest close rate?", he gets a written analysis grounded in the actual data — not a generic chatbot guess. The Operations Analysis report we generated in May surfaced that two of his four techs were crushing the $3K/day target while one was at 36% of peer revenue. That kind of signal isn't visible in a spreadsheet.

The custom connector to HouseCall Pro is where the moat lives. There's no off-the-shelf HCP integration for AI agents. We built one — and that work later became the foundation for HomeOps, the productized plugin we now install on any home services shop in 60 seconds.

HouseCall Pro APIDaily ops digest (6am CT)Weekly team reportTroubleshoots-over-timeRevenue-per-tech trackingClaude-powered analysisCustom HCP connector

Next.js · Vercel

Dashboard · daily digest

HouseCall Pro API

Custom connector

Claude · Sonnet

Analysis layer

Postgres

Historical metrics · trends

Resend

Scheduled report delivery

CallRail (planned)

Call review + coaching

04

Execution

The retainer is the part of this story that's actually replicable. AI tooling moves too fast for "we built it, here's your invoice" to make sense for any operator who wants the benefit of the platform over time. By the time the project would be billed and closed, half of what was possible six months later wouldn't be in the build.

So we structured Radon1 as an ongoing partnership at $1,900/month. The retainer covers dashboard operations, new agent capabilities scoped against the business, and continuous evaluation of new Claude models, new Anthropic features, and new integration possibilities as they ship.

The relationship is also a service exchange — Patrick's radon work meaningfully offsets the AI work. That kept the deal warm through the early months and made it possible to keep expanding scope without renegotiating every quarter.

~500

Hours/year saved

6:00 AM

Daily report drop time

3

Markets covered

$1,900

/mo managed retainer

05

Timeline

Feb 2026

Demo + proposal

Live demo of a general-purpose Claude agent in Slack. Conversation reframed: not 'use a $20 chatbot,' but build a dedicated operator connected to HouseCall Pro, CallRail, and Google Workspace. Proposal landed with two agents scoped and a managed retainer model attached.

Mar 2026

HCP integration + dashboard v1

Custom HouseCall Pro connector built. Dashboard live at reports.radon1.com. Daily 6am Central digest firing into Patrick's inbox. Weekly team report deployed. Real revenue numbers landing every morning instead of waiting on Ben.

Apr–May

Analysis layer + ops insights

Claude-powered operations analysis on top of the dashboard data. May 8 report surfaces a 64% revenue gap between top and bottom performers — actionable signal a static dashboard can't produce.

Ongoing

Managed AI expansion

Field Tech Agent and BI Agent scoped against Radon1's standards docs and operational data. Productized version (HomeOps) emerges from this engagement — now distributable to any HouseCall Pro shop. Retainer continues as the platform-side of the relationship.

06

What it proved

Custom connectors beat generic chat

There's no off-the-shelf HouseCall Pro integration for an AI agent. Building one turned Claude from a $20/mo curiosity into a system that knows Radon1's actual numbers, techs, and pipeline. That gap is the difference between 'interesting demo' and 'I check this every morning.'

Analysis is the new dashboard

Static charts don't tell an operator what to do. A Claude-powered report that says 'two of your four techs are 64% behind the leader' turns the same data into a decision. The analysis layer is where the leverage compounds.

Retainers fit AI delivery better than projects

The models change monthly. Anthropic ships new tooling weekly. A project-based engagement would have shipped a snapshot that decayed in value the moment it was delivered. A retainer keeps the system on the leading edge — and the operator doesn't have to follow the space to benefit from it.

One client engagement, one productized output

The HouseCall Pro work for Radon1 became the foundation for HomeOps — a distributable plugin that installs the same intelligence on any home services shop in 60 seconds. The retainer paid for the build. The product is the compounding return.

Running your ops on manual reports and spreadsheets?