← Back to Signal Labs AI Tooling Automations App Development

Meridian Health cut patient intake time by 68% with AI-powered workflows

A mid-market healthcare SaaS platform needed to modernize its intake and triage system. We built an AI-assisted workflow engine that reduced manual processing from 22 minutes to 7.

app.meridianhealth.io/workflows
Dashboard
Workflows
Patients
AI Triage
Reports
Settings
Workflow Performance
+ New Workflow
Avg. Intake Time
7.2 min
↓ 68% from baseline
AI Triage Accuracy
94.3%
↑ 12% since launch
Workflows Active
38
+14 this quarter
Intake Volume & Processing Time (12 weeks)
68%
Reduction in intake time
94%
AI triage accuracy
38
Automated workflows
12 wk
Idea to production

Manual processes slowing down patient care

Meridian Health's platform served 200+ clinics, but their patient intake system was still heavily manual. Staff spent an average of 22 minutes per patient on data entry, form routing, and initial triage. The existing system had no intelligence: every patient followed the same linear process regardless of urgency or complexity.

Their engineering team had explored off-the-shelf automation tools but found them too rigid for healthcare's compliance requirements. They needed something custom, built for their specific workflows, and smart enough to learn from their data.

AI-assisted triage with custom workflow automation

We started with a two-week discovery sprint. Shadowed clinic staff, mapped every touchpoint in the intake flow, and identified the three highest-impact bottlenecks: form routing, initial triage classification, and insurance pre-verification.

Rather than rebuilding the entire system, we designed an intelligent layer that sits on top of their existing infrastructure. The AI triage model classifies incoming patients by urgency and routes them through the appropriate workflow automatically. Staff only intervene when the model flags uncertainty.

1
Patient submits intake form
2
AI extracts and validates data
3
Triage model assigns urgency
4
Auto-routes to correct workflow
5
Insurance pre-verified via API
6
Staff reviews flagged cases only
Simplified view of the AI-assisted intake pipeline

Code-first prototyping, then production

We prototyped the triage model in the first sprint using a sample of Meridian's anonymized patient data. The working prototype was in front of clinic staff by week three for usability testing and accuracy validation.

The production build took eight additional weeks. We integrated with their existing EHR system, built a Power Automate layer for the simpler routing tasks, and developed a custom ML pipeline for the triage classification that improves with each patient interaction.

The workflow engine was built in React with a Node.js backend, deployed on their existing Azure infrastructure. The AI components use a fine-tuned model hosted on Azure OpenAI to stay within their compliance boundary.

React Node.js Azure OpenAI Power Automate PostgreSQL FHIR API Azure Functions
"We'd been trying to solve this internally for over a year. Signal Labs came in, understood our constraints immediately, and built something our team actually wants to use. The intake time reduction alone justified the entire engagement."
Dr. Sarah Chen, Chief Medical Officer, Meridian Health

From 22 minutes to 7, and getting faster

Within the first month of deployment, average intake time dropped from 22 minutes to 7.2 minutes. The AI triage model launched at 87% accuracy and climbed to 94.3% within three months as it learned from staff corrections.

Meridian has since expanded the workflow engine to cover 38 different process automations across their platform, from appointment scheduling to lab result routing. The system processes over 3,000 patient interactions daily across their network.

AI Tooling
Automations
App Development

Have a similar challenge?

Let's talk about how we can help you find signal in the noise.

Start a conversation