- Client
- ReSleuth / NationsRealty (Brad Stave)
- Service
- Full-Stack SaaS Development + DevOps Infrastructure
- Duration
- 47 hours concept to production; May–June 2026
- Key Result
- Production-grade lead management platform with MLS ingestion, AI-powered SMS outreach, and visual workflow designer — now the canonical CI/CD pattern for 11 PraeSyn-managed projects
The Challenge
ReSleuth provides predictive lead generation for real estate agents, using life-event analytics and behavioral signals to identify homeowners likely to sell. The product needed a complete lead management backend — but had no operational infrastructure at all.
Three core problems needed solving simultaneously:
1. MLS data ingestion and processing. Northwest MLS listing data needed to be pulled, parsed, and scored to identify high-value expired listings — homes where the listing contract expired without a sale, indicating a motivated seller.
2. Automated multi-step outreach at scale. Cold outreach to expired-listing homeowners requires careful orchestration: initial SMS contact, AI-based response classification, conditional follow-up branching, and agent handoff when a lead shows intent. Manual follow-up doesn't scale.
3. No production DevOps pipeline. The existing deployment process was manual — SSH into a VPS, run Docker Compose, pray. No CI/CD, no blue-green deploys, no health checks, no environment separation between dev and production.
Our Approach
PraeSyn built and deployed a complete lead management platform from scratch:
Lead Manager Application
- Stack: Next.js 16 + Prisma ORM + PostgreSQL + Twilio SMS
- Workflow Engine: Visual drag-and-drop workflow designer (workflow.io) — non-technical staff can build and modify lead engagement flows using configurable blocks: Trigger (New Listing), SMS Action, AI Decision branching, AI Memory read/write, Condition blocks, and Workflow Call chaining
- SMS Architecture: BlockExecutor pattern with 13/13 passing test suite — handles Twilio message delivery, DNC list filtering, template variable substitution, and response webhook handling
- MLS Integration: Complete field specification with data gap analysis for Northwest MLS, including Photos/CloudFront URL documentation
- Lead Dashboard: Web interface for pipeline visualization, lead detail views, and agent territory management
DevOps Infrastructure
- CI/CD Pipeline: First project fully migrated from Bitbucket Pipelines to the self-hosted Gitea + Woodpecker + Nomad stack — became the reference pattern for 11 subsequent project migrations
- Environments: Develop branch → leads.praesyn.com (dev), main branch → leads.resleuth.com (prod)
- Deploy Pattern: 3-step Woodpecker pipeline — Docker build → pre-pull image to VPS → Nomad job run with health check verification
AI-Powered Outreach
- AI Decision block classifies SMS responses (interested / not interested / wrong number / DNC request) and routes to appropriate follow-up
- AI Memory persists lead-scoped values across the engagement lifecycle
- Automated DNC filtering prevents compliance violations
The Results
Speed to market: ~47 hours of concentrated development delivered a production-grade lead management system — MLS data ingestion, SMS outreach engine, visual workflow designer, and full CI/CD pipeline all operational.
Quality assurance: 13/13 SMS BlockExecutor tests passing with verified outbound SMS pipeline. Root cause identified for initial delivery failures and resolved.
DevOps blueprint: The RELM deployment pattern — Gitea + Woodpecker + Nomad with 3-step pipeline — became the canonical template replicated across 11 other PraeSyn-managed projects.
Non-technical enablement: The visual workflow designer means staff without engineering backgrounds can build and modify lead engagement flows. Operations don't depend on developer availability.
Compliance built in: Automated Do Not Call filtering prevents compliance violations at the platform level — no manual list checking required.
What's Next
RELM is in active production at leads.resleuth.com. Ongoing work includes SMS provider redundancy for carrier diversity, statistical lead scoring models to improve conversion prediction, and expanded MLS field coverage as new data sources come online.