AI Automation Engineer for Remodeler

About us We're a Northern Virginia home-improvement group running multiple brands — kitchen & bath remodeling and exterior remodeling (roofing, siding, decks, windows, doors). We move fast, sell and produce a high volume of jobs, and we're drowning in manual coordination: lead follow-up, scheduling, purchasing, vendor data, and reporting. We want one operator to design and build an AI agent system that runs these workflows for us so the team can focus on selling and building. Our current stack: JobTread (CRM/project management), QuickBooks Online (accounting), Google Workspace (Gmail + Google Calendar), a WordPress website with lead forms, and recurring purchasing through vendors like Ferguson and Home Depot. What we're hiring you to build We've broken the project into modules. You don't have to build all of them at once — we expect to phase this and start with the highest-impact pieces. In your proposal, tell us how you'd sequence it. 1. Speed-to-lead & after-hours capture Auto-call (or text) a client back within minutes of a lead form being submitted on our website. Catch leads that come in when we're closed, qualify them, and book an estimate on the calendar automatically. Send the client a confirmation email and text as soon as the appointment is booked. Send the client a reminder text once the sales rep hits the road / is en route. 2. AI phone reception Screen inbound calls and politely refuse / filter out marketing and solicitation calls. Handle delivery-scheduling calls: confirm the delivery time and add it to our calendar. 3. Sales-team oversight Read through all emails from our sales reps and verify every customer is responded to in a timely manner — flag anything that's gone unanswered or is overdue. Read through the calendar and create a warning/reminder to place the material order one week after a job is sold. 4. Purchasing automation & vendor integration After an order is placed, connect to the purchasing email inbox, track the material, and give us a daily status update. Integrate with vendor sites such as Ferguson and Home Depot — pull product selections and order data automatically. Normalize PO/customer naming so everything is consistent: all PO names formatted equal and in capital letters so we can filter them. Reps are inconsistent — some enter a first name, some a last name, some only an address. The system must match a client by first name, last name, OR address and tie those records together (dedupe + identity matching). Goal: dramatically increase our organization and overall efficiency. 5. Contract → purchasing list → profit & loss Read each signed contract and automatically generate a purchasing list with expected purchasing cost. Capture actual cost as it comes in and reconcile it against expected. Produce a clean one-page expense sheet per job for profit & loss calculation. 6. Estimating & pricing knowledge assistant Learn our pricing metrics and answer pricing questions from the team accurately. Learn our proposal template and draft estimate scope-of-work documents in our format. 7. Bills & reminders Track and remind us about upcoming recurring payments — rent, payroll, and other bills. 8. Reporting cadence Every Monday: a brief of the day. Every Friday: a weekly summary plus a prioritized action list. Tools you'll likely work with You choose the stack, but here's where most of this lands. Tell us what you'd use and why. Voice / telephony AI: AI receptionist + outbound callback (e.g., Twilio, Vapi, Bland, Retell, ElevenLabs). Workflow automation: Make.com, n8n, or Zapier. LLM / agent orchestration: OpenAI / Anthropic APIs and an agent framework for the reasoning steps (email triage, contract parsing, scope writing, reporting). Document intelligence: OCR + LLM extraction to turn contracts into structured purchasing lists. Vendor data: API where available, secure browser automation / RPA (e.g., Playwright) where it isn't. Note: Ferguson and Home Depot don't offer open public APIs to most contractors — tell us how you'd reliably pull this data and keep it stable. Our systems: JobTread, QuickBooks Online, Google Calendar, Gmail, SMS. Who you are You've built production AI agents and multi-step automations that actually run a business — not just demos. Please show examples. Comfortable with telephony/voice AI, email and calendar APIs, web scraping/RPA, and LLM-based document extraction. You think about data hygiene and identity matching (fuzzy matching, deduping messy human input) as a first-class problem, not an afterthought. You design for reliability: error handling, logging, human-in-the-loop checkpoints, and clear daily output we can trust. You care about security — this system touches customer PII, contracts, and financials. Nice to have Experience in home services, construction, or contracting operations. Familiarity with JobTread and QuickBooks Online. Ability to document and hand off so our team can maintain it. How we'll work We'd like to start with a short discovery/scoping call, then build in phases — proving out one or two high-impact modules first (likely speed-to-lead + purchasing/PO normalization), then expanding. We value clear weekly communication and working software over big-bang delivery. Deliverables A working, documented agent system covering the modules above (phased). Integration setup for our existing tools. Documentation + a short training/handoff so our team can operate and adjust it. To apply, please include 2–3 relevant builds you've shipped (especially voice AI, purchasing/ops automation, or document-to-data extraction). Your proposed stack and how you'd integrate Ferguson / Home Depot reliably. How you'd phase the project and which module you'd start with. A ballpark timeline and cost (project and/or monthly retainer). Your availability and EST overlap.