Manage between-session check-in messages and triage responses before next appointment
Clinicians stop drowning in unread messages and start every session fully prepped with prioritized patient updates. This gives you a highly sticky, life-saving service to pitch to mental health practices that desperately need to reduce provider burnout and liability.
The problem today
15 hours
wasted weekly manually reviewing patient messages
40%
of critical patient updates missed before sessions
Marcus Chen is the owner and clinical director of a 5-therapist group practice in suburban Atlanta, managing roughly 140 active patients across his team. He keeps his phone on his nightstand every night because he knows that somewhere in his practice's message backlog, there could be a patient in crisis — and he has no system to tell him which one.
01The Problem
Every Monday, five clinicians read every weekend message blind — no prioritization, no triage, no way to start with what matters.
A patient moving from stable to crisis between sessions generates no signal until the situation has already escalated.
Clinicians enter each appointment cold, burning billable time reconstructing context that should have been ready before the door opened.
At this caseload size, structured between-session outreach is impossible — high-risk patients go dark between appointments with no detection.
A missed crisis signal buried in a routine inbox queue is a direct path to a licensing board complaint and malpractice exposure.
Front desk staff without clinical training can't flag distress in a message, so urgency is discovered by chance rather than by design.
02The Solution
Solution Brief
Fictional portrayal · illustrative
- Marcus runs 5 clinicians, 140 active patients, zero triage system
- Weekend messages read blind every Monday — all 140, in order
- Between-session check-ins skipped; caseload makes manual outreach impossible
- 48-hour window where patient deterioration is invisible to the practice
- Crisis missed in inbox = licensing complaint, not a corrected process
- First 10 minutes of every session lost to reorientation that structured prep would eliminate
- Marcus carries the clinical and legal risk of a gap his current tools cannot close
- Structured check-in sent to each patient 24–48 hours post-session
- Every response read on arrival; urgency dashboard ready before first appointment
- Crisis-level signals trigger immediate clinician alert and direct 988 Lifeline access — minutes, not morning
- Routine messages summarized; at-risk patients flagged without clinician inbox review
- HIPAA compliance, EHR integration, and crisis-safety logic make this infrastructure Marcus defends when budgets tighten
“I used to start every Monday by reading through 40-something messages with my stomach in a knot, just hoping I wasn't about to find out something bad happened over the weekend. Now the dashboard tells me exactly who needs my attention before I've had my coffee. I actually feel like I'm running a practice instead of just reacting to it.”
— Marcus Chen is the owner and clinical director of a 5-therapist group practice in suburban Atlanta, managing roughly 140 active patients across his team
03What the AI Actually Does
Automated Check-In Engine
Sends structured, clinically-informed check-in prompts to every patient 24–48 hours after their session via HIPAA-compliant SMS — no staff time required. Captures free-text patient responses and feeds them into the triage pipeline automatically.
Clinical Urgency Triage
Reads every patient response in natural language and sorts it into one of four urgency tiers: Routine, Needs Attention, Urgent, or Crisis. Clinicians stop reading every message and start acting on the ones that matter.
Crisis Detection & Escalation
Scans incoming messages for crisis signals — suicidal ideation, acute distress patterns, sudden behavioral shifts — and triggers an immediate clinician alert while simultaneously delivering 988 Suicide & Crisis Lifeline information directly to the patient. Measured in minutes, not hours.
Pre-Session Briefing Dashboard
Delivers a prioritized summary of every patient's between-session status to the clinician before their first appointment of the day. Clinicians walk in knowing who had a hard week, what changed, and where to focus — instead of spending the first ten minutes of a session catching up.
04Technology Stack
OpenAI API (GPT-4.1-mini) — Primary LLM
$0.40/1M input tokens, $1.60/1M output tokens. Estimated $50–$150/month for 200 patients. MSP resale: $100–$250/month
Primary LLM for processing patient check-in responses, sentiment analysis, triage classification, and generating structured pre-appointment summaries.…
OpenAI API (GPT-5.4) — Escalation LLM
$2.50/1M input tokens, $10.00/1M output tokens. Estimated $20–$80/month for escalated messages only. MSP resale: $40–$140/month
Higher-capability LLM used only for messages flagged as clinically significant by the primary model or crisis detection layer. Provides deeper reasoni…
Twilio Programmable Messaging (SMS)
$0.0079/SMS sent or received + $2/month per phone number. Estimated $30–$100/month for 200 patients. MSP resale: $75–$200/month
HIPAA-eligible SMS delivery platform for sending check-in prompts to patients and receiving their responses. Twilio signs BAAs for eligible products. …
Healthie EHR/Practice Management Platform
$149.99/month (Group Plan, 1 provider + 1 support) + $50/month per additional provider. MSP resale: $250–$400/month
Recommended EHR with full GraphQL API and webhook support. Provides the appointment schedule data (to trigger check-ins), patient roster, and the abil…
AWS ECS Fargate (Container Compute)
$50–$150/month for agent containers (2 vCPU, 4GB RAM typical). MSP resale: $100–$300/month
Serverless container hosting for the AI agent core service, API gateway, and clinician dashboard backend. No server management required. Runs within H…
AWS RDS for PostgreSQL (Database)
$30–$80/month (db.t3.medium, 20GB storage, encrypted). MSP resale: $60–$160/month
Managed PostgreSQL database for storing patient check-in conversations, triage results, clinician configurations, audit logs, and session summaries. E…
AWS SQS (Message Queue)
$1–$5/month at typical volume. Included in cloud hosting markup.
Asynchronous message queue for decoupling inbound patient SMS reception from AI processing. Ensures no messages are lost during processing spikes and …
AWS S3 (Encrypted Object Storage)
$5–$15/month. Included in cloud hosting markup.
Encrypted storage for conversation archives, audit logs, consent records, and system backups. Server-side encryption with AWS KMS. Required for HIPAA …
AWS KMS (Key Management Service)
$1–$3/month per key + $0.03/10K requests. Included in cloud hosting markup.
Centralized encryption key management for all data at rest (RDS, S3) and application-level encryption of PHI fields. Customer-managed keys provide ful…
n8n (Self-Hosted Workflow Automation)
$0 license cost; runs on existing AWS infrastructure. MSP labor for setup included in implementation.
Workflow orchestration engine that connects the appointment schedule polling, check-in message dispatch, inbound message routing, LLM processing pipel…
Auth0 (Identity & Access Management)
Free tier for up to 7,500 MAU; $23/month for Professional. MSP resale: $50–$100/month
Provides SSO, MFA, and role-based access control for the clinician dashboard. Supports HIPAA-eligible configurations with audit logging of all authent…
05Alternative Approaches
Pre-Built Mental Health Chatbot Platform (Wysa or Woebot)
Wysa: $74.99/year individual; custom enterprise pricing. Woebot: custom enterprise pricing.
Instead of building a custom AI agent, license an existing clinical AI chatbot platform like Wysa or Woebot. These platforms offer pre-built, clinically validated conversational AI for mental health check-ins with existing evidence base and FDA considerations. The MSP would handle integration and IT management rather than full custom development.
Strengths
- Faster deployment (weeks vs. months)
- Clinically validated content
- Existing evidence base
- Lower development cost
- Reduced clinical safety engineering burden
Tradeoffs
- No clinician-facing triage dashboard (major gap)
- Limited EHR integration
- Cannot customize check-in domains per clinician
- No pre-appointment summary generation
- No direct clinician alerting
- Practice has less control over the experience
- Ongoing per-user licensing costs may exceed custom solution at scale
Best for: Practice wants basic patient engagement without custom triage/dashboard, budget is under $20K for implementation, or timeline is under 4 weeks.
Keragon + StackAI No-Code Approach
Keragon: $49–$399/month; StackAI: separate pricing. Combined platform costs: $200–$600/month.
Use Keragon for healthcare-specific workflow automation combined with StackAI for no-code HIPAA-compliant chatbot building. Both platforms are HIPAA-ready with BAAs included. Check-in flows, LLM integration, and EHR connections are configured via visual drag-and-drop interfaces. The clinician dashboard would use Keragon's built-in notification features plus a simple portal.
Strengths
- 60–80% reduction in development time
- No custom code to maintain
- HIPAA compliance built in
- Lower MSP technical expertise required
- Easier to modify workflows post-deployment
Tradeoffs
- Less control over crisis detection logic (must work within platform constraints)
- Potential vendor lock-in to Keragon/StackAI
- Monthly platform costs add up ($200–$600/month for platforms alone)
- Clinician dashboard less customizable
- May hit platform limitations as requirements evolve
Best for: MSP lacks deep Python/AI engineering talent, timeline is 6–10 weeks, practice is comfortable with SaaS dependency, or this is the first deployment before investing in a custom platform.
Spruce Health as Primary Platform
$24/user/month
Use Spruce Health as both the secure messaging platform and the clinician communication hub, with a lightweight AI processing layer that intercepts Spruce messages via API. Spruce provides HIPAA-compliant messaging, phone, video, and team collaboration already used by many mental health practices. The AI agent would be a thin service that monitors Spruce conversations and injects triage metadata.
Strengths
- Spruce is already adopted by many mental health practices (minimal behavior change)
- Built-in HIPAA compliance
- Clinician already has a familiar interface
- Lower cost for messaging infrastructure ($24/user vs. building custom dashboard + Twilio)
Tradeoffs
- Less control over the patient messaging experience
- Spruce API access may be limited for custom AI integration
- No appointment-triggered automation (manual check-in initiation)
- Pre-appointment summaries would need to be pushed as Spruce messages rather than EHR notes
Best for: Practice already uses Spruce, clinicians resist adopting a new dashboard, budget is constrained, or the practice wants incremental AI augmentation of existing workflows rather than a new system.
Azure OpenAI + Azure Stack (Single Vendor)
Comparable to AWS stack; slightly higher infrastructure costs for equivalent compute.
Deploy entirely on Microsoft Azure using Azure OpenAI Service for LLM inference, Azure App Service for the agent, Azure Database for PostgreSQL, Azure Communication Services for SMS, and Azure AD B2C for authentication. This consolidates all vendors under a single Microsoft BAA and Azure HIPAA compliance umbrella.
Strengths
- Single BAA covers all infrastructure and AI services
- Simplified compliance posture
- Azure OpenAI provides same GPT models with Microsoft's enterprise security wrapper
- Azure Communication Services is HIPAA-eligible for SMS
- Strong enterprise support
Tradeoffs
- Azure OpenAI may have higher latency for API calls
- Azure Communication Services is less feature-rich than Twilio for SMS
- Azure AD B2C is more complex to configure than Auth0
- Slightly higher infrastructure costs for equivalent compute
- Less flexibility to switch LLM providers
Best for: Practice or MSP is already an Azure shop, compliance team prefers single-vendor approach, enterprise Microsoft agreement is in place, or the MSP has stronger Azure than AWS expertise.
Manual Triage with AI-Assisted Summaries Only
Lower development and ongoing costs due to eliminated triage engineering; AI summary generation costs only.
Deploy a simplified system that sends check-in messages and collects responses, but skips automated triage entirely. All patient responses are displayed chronologically in the clinician dashboard for manual review. The AI only generates pre-appointment summaries from the raw conversation data. No autonomous crisis detection — clinicians review all messages themselves.
Strengths
- Dramatically simpler to build and validate (eliminates the entire crisis detection and triage engineering effort)
- Lower clinical risk (no AI making triage decisions)
- Faster deployment (8–12 weeks)
- Lower ongoing maintenance
- No risk of missed AI crisis detection
Tradeoffs
- Clinicians must review ALL messages (higher time burden)
- No automated prioritization (urgent messages may be buried)
- No real-time crisis alerts (delays depend on when clinician checks dashboard)
- Loses the core value proposition of autonomous triage
Best for: Practice is uncomfortable with AI making clinical triage decisions, regulatory environment is uncertain, practice has fewer than 50 active patients making manual review feasible, or as a Phase 1 deployment before adding AI triage in Phase 2.
Ready to build this?