How 4,000 Hospitals Got On-Demand Video Interpreters in Under 20 Seconds — Across 250 Languages
The Problem
Equiti Health's Martti platform (My Accessible Real-Time Trusted Interpreter) — the first video remote interpreting service built for healthcare — needed to scale from a pioneering VRI product into the infrastructure thousands of US hospitals depend on for language access. The engineering challenge: build a WebRTC platform that matches a patient's language to an available interpreter and establishes clinical-quality video in under 20 seconds, integrates natively into hospital EHR workflows, runs across every device type in clinical environments, and maintains 99.99% uptime — all under HIPAA and SOC 2 compliance. Equiti Health had the domain expertise and the interpreter network; Trembit came on board as the dedicated engineering team to build and scale the platform.
Why This Doesn't Work With Off-the-Shelf Video
General-purpose video platforms connect known participants in scheduled calls. Medical interpretation connects an unknown language need to an available interpreter and establishes a clinical-quality session in under 20 seconds, inside a hospital EHR workflow — a fundamentally different engineering problem:
- Intelligent routing, not a meeting link — identify the patient's language (90+ via video, 250+ via audio, including ASL), find an available qualified interpreter, and establish an encrypted WebRTC connection, all within the sub-20-second window
- Native EHR and telehealth integration — one-click launch from Epic (Hyperspace, Hyperdrive, Haiku, Canto, Rover), Oracle Cerner, and eClinicalWorks with automatic language detection and session write-back, plus embedding into Zoom, Teams, Webex, Doxy.me, Amwell, Teladoc, and SecureVideo
- ASL-specific media engineering — sign language demands sustained resolution and frame rate that standard WebRTC compression actively degrades; a compression artifact during a clinical encounter is a patient-safety risk, not a quality nuisance
- Healthcare-grade infrastructure — encrypted, auditable media under HIPAA and SOC 2, Joint Commission documentation, and 99.99% uptime (52 minutes/year) across thousands of concurrent sessions on constrained hospital networks
What We Did
Core Video Infrastructure
- Designed and implemented WebRTC-based video remote interpreting (VRI) supporting 90+ languages via live video and 250+ languages via audio
- Built the intelligent routing algorithm that matches patient language to available interpreters and establishes clinical-quality video connections in under 20 seconds
- Engineered ASL support with 24/7 nationally RID-certified interpreters and specific video-quality standards, with encrypted media streams compliant with HIPAA and SOC 2 throughout the path
Platform, Integrations & Applications
- Built Epic EHR integration across all major modules (Hyperspace, Hyperdrive, Haiku, Canto, Rover) — one-click launch, automatic language detection, and session write-back — plus Oracle Cerner and eClinicalWorks with the same smart matching and auto-documentation
- Embedded interpreter access within Zoom, Microsoft Teams, Cisco Webex, Doxy.me, Amwell, Teladoc, and SecureVideo, and built native iOS (Swift) and Android (Kotlin) apps plus web access for bring-your-own-device deployment
- Built the management portal — heatmap-based demand forecasting, real-time session monitoring across all facilities, and in-house interpreter routing with Martti as automatic backup
Scale, Security & Reliability
- Achieved and maintained 99.99% uptime across 4,000+ healthcare facilities — the reliability standard required when interpreter access directly impacts patient outcomes
- Built comprehensive monitoring and alerting, segmented billing by facility and department, and supported SOC 2 (AICPA) certification with ongoing HIPAA compliance
- Designed a CI/CD pipeline for rapid deployment and security patching without service interruption
Key Results
In Their Words
Their processing integrity was impressive. They precisely estimated the number of hours required and delivered according to promises.
Their proactive team gets things done as if it were their own project.
What We Learned
Sub-20-second connections at scale require protocol-level WebRTC, not wrapper SDKs
Matching a patient's language to an available interpreter and establishing a clinical-quality connection in under 20 seconds — across 90+ video languages, thousands of concurrent sessions, and hundreds of device types — is not a configuration exercise on a hosted service. It requires building the routing intelligence, media infrastructure, and quality assurance layer from the protocol level up. If your platform depends on connection speed and reliability at scale, the WebRTC-layer decisions determine everything.
EHR integration is an architecture decision, not a feature addition
Integrating with Epic across five modules, Oracle Cerner, and eClinicalWorks — with one-click launch, smart language matching, and auto-documentation — requires the integration to be designed into the platform architecture from the start. Teams that build the core platform first and "add EHR later" end up rebuilding significant portions. If your platform will live inside clinical workflows, design for EHR integration before the first line of code.
99.99% uptime at healthcare scale requires specific infrastructure patterns
52 minutes of annual downtime across 4,000+ facilities is not achieved with standard cloud architecture. It requires redundancy at every layer, real-time monitoring with automated failover, zero-downtime deployment pipelines, and infrastructure that treats every session as life-critical — because in healthcare interpretation, it often is.
Building a Healthcare Platform at Scale?
Building a healthcare platform that needs to integrate with EHR systems and scale to hundreds of facilities? We have done this at the scale that matters — from Epic integration patterns to 99.99% uptime across 4,000 hospitals. Book a 30-minute architecture session. No pitch deck. Just engineering clarity.