AI DEV CASE STUDY
SWRS CRM: An AI-Powered Operations Platform
One platform for the entire home-improvement contractor workflow, from lead to payment, with AI built into the stages where teams lose the most time: scoping, inspection, insurance, lead scoring, and training.
Project Context
Client
SWRS (Statewide Roofing Specialist)
Niche
SaaS · AI CRM
Build
~1 year · 3-4 developers
The Tech Stack
- Node.js
- Express.js
- Prisma ORM
- PostgreSQL
- AWS RDS
- React.js
- AWS (S3 · SES · SNS · CloudFront)
- Docker
- PM2
- OpenAI
- ElevenLabs
- Telnyx
"We wanted a single, AI-powered platform that connects the entire contractor workflow, from the first lead through job, invoicing, and payment, and gives our teams intelligent tools to work more efficiently."
Contractors usually stitch this together from half a dozen disconnected tools. The real work was one platform that carries the whole workflow and puts AI exactly where the manual effort is.
Terry Cribb
Founder, SWRS
HOW WE BUILT IT
A Structured Build Around a Lead-to-Payment Workflow
Delivered over about a year by a small team, the project ran across a multi-organisation core, the lead-to-payment workflow, several AI integrations, voice and telephony, and a customer portal. We agreed the data model and workflow up front, then built outward from it in reviewable stages.
Discovery
A working session with the Oh Pretty team to understand the brand positioning, the dress range, the occasion-based selling model, and how the
team wanted to manage content going forward.
Requirement Analysis
Split the brief into workstreams: multi-org core, workflow stages, AI integrations, voice, homeowner portal, and permissions.
Multi-Org Architecture
Built the multi-organisation model so one platform serves different contractor businesses, teams, roles, and permissions with isolated data.
Core Workflow Build
Connected the whole chain: Lead → Inspection → Insurance → Proposal → Acceptance → Job → Invoice → Payment.
AI Integration
Wired OpenAI into lead scope generation, insurance document extraction, image and damage analysis, lead
scoring, and assessment evaluation.
Voice & Telephony
Integrated ElevenLabs conversational agents for mock-call training and Telnyx programmable voice for CRM calling.
Portal & Roles
Built the dedicated homeowner portal and the role-based access controls across the team.
QA & Launch
Tested every workflow stage, integration, and AI output across roles and devices before the platform went live.
AI Lead Scoring & Prioritisation
Every lead carries an AI score, a priority, and a plain-language reason, so sales teams work the highest-value opportunities first instead of guessing. The score blends conversion likelihood, urgency, project potential, source, and engagement.
AI Damage Detection
Inspection photos, over a hundred per inspection, run through OpenAI image analysis to flag potential roof damage by location and severity, feeding the estimate and proposal directly from what the inspector captured on site.
Insurance Scope Extraction
The hardest problem in the build: unstructured insurance scopes and documents are parsed by AI into structured line items, measurements, materials, quantities, and pricing that flow straight into the CRM and the proposal.
AI Proposals with Multiple Quotes
The proposal engine handles measurements, materials, services, upgrades, excluded work, discounts, and calculations, and presents multiple quotes, Basic, Standard, Premium, inside one proposal, keeping the accepted value aligned with the job and invoice that follow.
Lead-to-Job Project Pipeline
Each project moves through a clear multi-step pipeline, lead, appointment, inspection, contingency, proposal, job, final inspection, invoice, close, with the customer, payment, and current stage all visible in one place.
Calendar & Appointment Scheduling
An integrated calendar schedules, reschedules, assigns, and tracks sales and field appointments across the whole team, so nothing falls between an office manager and a crew in the field.
Invoicing & Payments
Invoices, payment schedules, and collection run inside the platform, so the money side stays tied to the exact job it came from rather than living in a separate accounting tool.
LMS & Employee Training
An integrated LMS delivers courses, modules, videos, and assessments with progress tracking, and AI scores the more complex assessments, going beyond simple multiple-choice marking.
AI Mock-Call Training
Employees practise real sales conversations against ElevenLabs voice agents; each call is evaluated by AI on communication, objection handling, product knowledge, and customer interaction, then scored back into the LMS.
WHAT TO EXPECT FROM US
Transparent Build Process and Clean Delivery for AI Platforms
If you are building an AI-powered platform and weighing partners, here is what working with our team actually looks like.
A plan before a proposal.
We start by mapping what your platform must do end to end, who uses it, and where the manual work is. You leave with a delivery plan, not a sales pitch.
One platform, not a plugin pile.
We connect the whole workflow in one system, rather than stitching together apps that each own a slice and none own the handoffs between them.
AI where it earns its place.
We add AI to the stages that genuinely cost your team time, scoping, extraction, scoring, evaluation, not as a demo feature bolted on top.
Clean, maintainable code.
We write Node and React that any competent engineer can read and extend, so you are not locked to us after launch.
Integrations built to run unattended.
OpenAI, ElevenLabs, and Telnyx are wired to keep working in production, with the failure cases handled, not just the happy path in a demo.
A milestone review at every stage.
You review the real system at each stage: architecture, workflow in staging, AI outputs, final QA. No month-long silence before a reveal.
Quality that does not depend on one person.
Every workflow stage and AI output is tested across roles and devices, not eyeballed once on a single screen.
You own what we build.
At handover you get the full stack: source, integration configuration, infrastructure, and documentation. Your team can run it independently or keep us on.