AI Client Reactivation Agent: Two-Way SMS + Email, HIPAA-Ready Multi-Tenant SaaS

I'm hiring a developer to build an AI agent that texts and emails a business's inactive clients and gets them to book again. Version 1 takes clients from a CSV upload, so it works with any client software. Launch market is healthcare (chiropractic, PT, dental, med spa and similar), piloting in my own clinic in Anchorage, AK. It will then be sold as SaaS to any service business (salons, home services, veterinary, legal and more). HOW IT WORKS 1. Staff upload a CSV (name, mobile, email, last visit, consent per channel). Columns are matched automatically. 2. The system picks clients by inactivity rules, consent, opt-outs and recent contact. 3. It sends a personalized text from the business's own Twilio number, or an email when the client has email consent only, using approved templates and only during allowed hours. 4. An LLM handles text replies within strict guardrails: scheduling and logistics only, never professional advice. 5. Each conversation ends with a booking link, a staff handoff alert, an opt-out, or one follow-up and then close. 6. A one-screen dashboard shows clients brought back, replies, handoffs and opt-outs. SIMPLICITY FIRST A front-desk person with no training must upload a list and launch a campaign in under 15 minutes, in no more than 3 screens. Industry presets fill in terminology, templates and timing, advanced settings are hidden, and compliance runs automatically. SCOPE (VERSION 1) - Multi-tenant from day one, with role-based access and a super-admin view - Swappable plug-ins for client source (CSV), channel (two-way SMS, one-way email; voice later) and booking (link, staff handoff) - Industry terminology, profiles and guardrails as configuration, not code; 3 starter profiles - Compliance in the core: consent records per channel, permanent cross-channel opt-outs, FCC Reassigned Numbers check, state quiet-hours and frequency rules, promotional-content check, append-only audit log, consent and message history export - HIPAA-grade security; every vendor touching client data signs a BAA (Twilio; AWS SES or Paubox; Claude via Anthropic or AWS Bedrock; AWS, GCP or Azure). Self-hosted n8n or a light Node/Python backend. OUT OF SCOPE: EHR/CRM integrations, writing into schedules, billing, voice (Phases 2 and 3; quotes welcome but optional), self-serve signup. MILESTONES AND ACCEPTANCE (fixed price, Upwork escrow, pilot in about 4-6 weeks) 1. Architecture and setup (15%): written design approved (plug-in interfaces, multi-tenant data model, industry configuration, hosting plan); all accounts in my name; BAAs confirmed. 2. Import, SMS and email core (30%): a sample CSV imports correctly; test numbers get texts and test inboxes get emails; STOP or unsubscribe opts a client out of every channel; sends outside quiet hours are held until the window opens. 3. AI conversation and handoff (30%): the AI handles 20 scripted test conversations correctly, refuses professional-advice questions and triggers staff alerts, in both a healthcare and a non-healthcare account. 4. Dashboard, security and pilot (25%): dashboard and audit log work; each account's data is isolated; documentation delivered; a first-time user launches a campaign in under 15 minutes; the pilot runs in my clinic for 2 weeks with no critical bugs. Also required: a send without consent is blocked; a promotional template to an informational-consent-only client is blocked; any test client's full consent and message history exports; a non-healthcare account runs end to end with no code changes. DELIVERABLES: source code in my private GitHub; production and staging in my cloud account; technical docs (setup, deployment, adding a plug-in or industry profile); admin onboarding guide; ~30-minute walkthrough video; 30-day bug-fix warranty. TERMS: work for hire with full IP assignment; all accounts in my name; BAA and NDA before any real client data (use synthetic data in development); only open-source licensed for commercial resale (MIT, Apache or similar); no lock-in to your personal accounts or private libraries. IDEAL CANDIDATE: has shipped production two-way SMS apps (Twilio preferred); has built LLM agents with guardrails and human handoff; has HIPAA experience (BAAs, PHI, audit logging); has built simple apps for non-technical users and multi-tenant SaaS. n8n and healthcare software experience are pluses. Clear English, weekly updates and a 30-minute weekly call. This could grow into ongoing work. The budget shown is a target; I'm open to proposals. Answer the screening questions, and start your proposal with the word "Spine" so I know you read the whole post.