We work where the rules shape the architecture

EHR integration, clinical workflow and the compliance work around both.

We build on top of the systems clinics actually run. That means HL7 v2 and FHIR interfaces, OpenEMR customization and migration, and the unglamorous work of reconciling patient records between two vendors who model them differently. Everything is designed for a HIPAA review from the first commit: audit trails, least-privilege access, encrypted at rest and in transit, and a BAA in place before any PHI moves.

• EHR and EMR integration (FHIR, HL7 v2, CCDA)
• OpenEMR customization, migration and support
• Practice management and clinical ERP builds
• Patient portals, intake and e-consent flows
• Claims, eligibility and revenue-cycle tooling
• HIPAA-aligned infrastructure and audit logging

• FHIR
• HL7 v2
• OpenEMR
• HIPAA
• Epic
• Cerner
• CCDA

9 EHR integrations shipped since 2021

Payments, ledgers and the reconciliation nobody demos but everybody needs.

Stripe and Adyen integrations, double-entry ledgers that actually balance, and the reconciliation jobs that catch it when they do not. We have built card-present and card-not-present flows, subscription billing with proration, marketplace payouts under Connect, and KYC/AML onboarding wired into third-party providers. Most of this work is regulated, so we plan for the audit rather than retrofitting it.

• Stripe, Adyen and Plaid integrations
• Double-entry ledger and reconciliation systems
• Subscription billing, proration and dunning
• Marketplace payouts and split payments
• KYC/AML onboarding and identity verification
• PCI-DSS scope reduction and tokenization

• Stripe
• Adyen
• Plaid
• PCI-DSS
• ACH
• KYC/AML
• Ledgers

$0 in reconciliation breaks across 3 live ledgers

LLM features that survive contact with real users and real invoices.

We are asked for a chatbot and usually deliver something narrower and more useful: a retrieval system over documents the business already owns, an agent that completes one workflow end to end, or a classifier that removes a queue of manual review. Built on LangChain and LangGraph, evaluated before launch, and instrumented so you can see quality and spend per feature rather than one number on a bill.

• Document retrieval and internal knowledge search
• Workflow agents built on LangGraph
• Customer-facing copilots with guardrails
• Document extraction and classification
• Evaluation suites and quality dashboards
• Model routing and cost optimization

• LangChain
• LangGraph
• RAG
• Agents
• Evals
• Vector search

Median 41% cut in LLM spend after evaluation work

Learning platforms, SIS integration and assessment at scale.

Course delivery, cohort management and the standards work that lets a platform talk to the systems a district or university already runs: LTI 1.3, SCORM, xAPI, OneRoster. We also handle the parts that get schools into trouble, including FERPA-aligned data handling, accessibility to WCAG 2.2 AA, and proctoring or assessment flows that hold up when three thousand students start at once.

• LMS and course-delivery platforms
• LTI 1.3, SCORM and xAPI integration
• SIS and OneRoster synchronization
• Assessment, grading and proctoring flows
• WCAG 2.2 AA accessibility remediation
• FERPA-aligned data architecture

• LTI 1.3
• SCORM
• xAPI
• OneRoster
• FERPA
• WCAG 2.2

Platforms serving 120k+ enrolled learners

Field data, equipment telemetry and forecasting that works offline.

Agtech has a constraint most software does not: the people using it are frequently out of signal. We build offline-first field applications that sync when they reconnect, ingest telemetry from equipment and sensors over MQTT, and put satellite and weather data behind yield and irrigation models. The reporting on top has to satisfy both an agronomist and a compliance auditor, which usually means two views of the same data rather than a compromise between them.

• Offline-first field data capture
• Equipment and sensor telemetry (MQTT, LoRaWAN)
• Satellite and weather data pipelines
• Yield, irrigation and input-planning models
• Traceability and compliance reporting
• Cooperative and grower portals

• MQTT
• LoRaWAN
• GIS
• NDVI
• Offline-first
• Telemetry

Field apps used across 340k acres

Perception tooling, fleet telemetry and the data loop behind both.

We work on the software around autonomous and connected vehicles rather than the vehicle itself: data ingestion from fleets, annotation and triage tooling for perception teams, simulation harnesses, and the pipelines that turn a week of driving logs into a training set. On the connected-vehicle side, telematics ingestion, OTA update orchestration and diagnostics dashboards for fleet operators.

• Perception data pipelines and annotation tooling
• Scenario extraction and simulation harnesses
• Fleet telematics ingestion and dashboards
• OTA update orchestration
• ROS 2 tooling and log triage
• Driver-behavior and diagnostics analytics

• ROS 2
• Telematics
• OTA
• Perception
• Simulation
• CAN bus

2.1PB of driving logs processed to date