Does Interbox provide out-of-the-box integrations?

No, and neither can anyone else. In health IT there's no drop-in integration: every sender speaks its own dialect of HL7v2 with its own local codes, so no honest tool can auto-map your data — anything that claims to is guessing. Interbox gives you a head start instead: ready-made templates and worked examples for common sources and formats, so you adapt a proven mapping to your feed rather than starting from a blank file.

If Interbox is built around the mapping-as-code idea, will I be on my own when building mappings?

No. Mapping-as-code means you own the result, not that you're on your own. The SDK ships typed building blocks for HL7v2, X12, C-CDA and FHIR, the console shows unmapped messages and drafts AI fixes you review before shipping, and our engineers can build, deploy, manage and support your mappings for you — with the code still in your repo and on your infrastructure.

Does Interbox work with any FHIR server?

Yes. Interbox delivers FHIR transaction Bundles to any conformant FHIR server, with native optimized delivery to Aidbox.

What data formats does it support?

HL7v2, X12, C-CDA, and FHIR out of the box, plus custom formats via pluggable sources.

Can I run it against PHI?

Yes. The Standalone tier is licensed for production PHI and runs in your environment; the free Developer tier is for synthetic data only.

Can I extend it with my own connectors?

Yes. Built-in stages ship in the SDK, and you can write your own sources, mappers, and senders in your repository.

How do I test my mappings?

Mappers are TypeScript with unit tests in your repo, run in CI; you can also replay messages through the console before shipping.

How does terminology mapping work?

Concept maps translate local codes to LOINC, SNOMED and FHIR value sets; unmapped codes are queued in the console and resolved without a redeploy.

How is Interbox deployed?

Deploy Interbox to any Kubernetes cloud with Helm charts or on-premise; it ships as a container with a PostgreSQL database, running alongside your FHIR server.

Healthcare integration engine

Interbox

Own your integrations like the rest of your stack — reviewed in PRs, tested in CI, versioned in your own repo.

Every message is traceable end to end, and when one fails the AI assistant drafts the mapping, so the fix ships as a diff you review.

Try now Docs Book a demo

01 · Observe

Be confident in your pipelines

One screen for ingestion, mapping and delivery — live throughput, queue depth at every stage, and the health of Postgres, Aidbox and the engine. Nothing is dropped; every message is accounted for.

Real-time throughput and error rate
Queue depth for received, processed, sent, error and deferred
Postgres, Aidbox and engine health at a glance

Ingestion

24msg/s

+6%

Mapper

23msg/s

+5%

Sender → Aidbox

23msg/s

+4%

Error rate

0.4%

-0.1%

Throughput · 60s

403020100

-59s-44s-29s-14s-0s

inboundmappedsenterrors

Queue depths

received412

processed96

sent84.6k

error23

deferred11

active partition542 rows

Top error kinds · 24hView all

unmapped_code	127
validation_error	48
mapping_error	31
prerequisite_missing	19
aidbox_5xx	6

System health

Postgres12/100 conns · 4.2 GB3ms

Aidboxyour-org.aidbox.app8ms

Enginethreads 3 (1·1·1) · up 6d 4h18% cpu

Mapping coverage137 unresolved94%

Workspaceyour-org/integrationsremote

02 · Search

Find any message in seconds

Facet by status, source and format, then drill into any failure to see where it stopped and why. A message flagged error keeps its cause attached; one that's deferred is parked to revisit — you retry on your own schedule.

Faceted search across status, source, format and error kind
Query bar for ad-hoc filters
Open the raw payload and inspect its FHIR output

status:error source:ed-system⌘K

Facetsclear

status

received412

processed96

error23

deferred11

source

Hospital EHR600

Clinic EHR320

ED system180

Reference lab100

format

HL7v2560

X12360

C-CDA180

FHIR100

error_kind

unmapped_code127

validation_error48

mapping_error31

23 results

message	status	received	format	source	patient
0038	error	9s	ADT^A08	ED system	Walker, Tom
0033	error	22s	Claim 837	Clinic EHR	Nakamura, Yui
0026	error	44s	ORU^R01	Reference lab	Brown, Sara
0021	error	1m	C-CDA	Hospital EHR	Müller, Hans
0018	error	1m	ADT^A03	ED system	Patel, Aisha
0012	error	2m	Bundle	Registry	Lee, Chen
0007	error	3m	ADT^A08	Hospital EHR	Okafor, Noah

03 · Resolve

Build and expand your local code dictionary

Unmapped local codes are caught and queued instead of dropped. Pick a target — LOINC, SNOMED, a FHIR value set — approve it, and reprocess the parked messages. The built-in AI assistant reads the failing message and proposes the fix for you.

Concept maps translate local codes to standard terminologies
Approve a target and reprocess deferred messages in place
AI assistant diagnoses failures and suggests the mapping

Local Dx → SNOMED CTHL7v2

91 open

Lab codes → LOINCX12

28 open

Race → OMB CDCC-CDA

12 open

Facility → LocationFHIR

6 open

Local Dx → SNOMED CTHL7v2

ConceptMap/local-dx-snomed

Export Edit metadata

Sending application

HOSPITAL-EHR

Sending facility

MAIN

Mapped

1389 (94%)

Open

Unmapped 91Mapped 1389FHIR resource

91 unmapped codesPick a target, then Approve.

Local code	Display	Target	Messages
DM-T2-UC	Type 2 DM, uncontrolled	Pick target…	18	Pick target
ANEMIA-CX	Anemia, unspecified	271737000 Anaemia	12	Approve
CP-ATYP	Atypical chest pain	Pick target…	9	Pick target
HTN-2	Essential hypertension	59621000 Hypertensive disorder	7	Approve
UTI-AC	Acute cystitis	68566005 UTI	5	Approve
COPD-EX	COPD exacerbation	Pick target…	4	Pick target

Assistant claude-opus-4

6 messages in context

Why are inbound ADT^A08 messages failing?

23 failed with mapping_error — the facility code "CLINIC-77" isn't in facility-location. Map it to a Location and the parked rows can be retried.

Resolve it.

All 23 carry it in PV1-3.4. Closest match in Aidbox is Riverside Family Clinic — NPI 1487654321, same sending application. Proposed entry:

source CLINIC-77

target Location/riverside-fc

equiv equivalent

Approve it and reprocess.

Entry written to ConceptMap/facility-location. 23 deferred messages requeued — 23 sent, 0 errors.

Message the assistant… / commands

04 · Own

Your mappings are code you own

Every converter is versioned TypeScript in your own Git repo — reviewed in PRs, tested in CI, hot-reloaded the moment you push. No proprietary DSL, no black-box visual mapper, and you can always write your own source or sender.

Push to your branch, pull from the console, hot-reload
Unit tests next to your mappers, run in CI
AI assistance and format specs embedded in the console

main a3f1c92 · Tanya Volkov · 8m ago hot-reload active Pull now Open on GitHub

Code Commits 7 Diff 2

integrations18 files

src/
├── index.ts
├── pipelines/
│   ├── index.ts
│   └── intake-to-aidbox.ts
└── mappers/to-fhir/
    ├── index.ts
    ├── patient.ts
    ├── encounter.ts
    ├── observation.ts
    ├── coverage.ts
    ├── condition.ts
    ├── datatypes/
    │   ├── humanname.ts
    │   ├── identifier.ts
    │   └── address.ts
    └── support/
        └── datetime.ts
test/
└── mappers.test.ts
package.json

mappers/to-fhir/patient.tsa3f1c92 VS Code

// PID → FHIR Patient
import { PID } from "@healthsamurai/interbox/hl7v2";
import { toIdentifier } from "../datatypes/cx-identifier";
import { toHumanName } from "../datatypes/xpn-humanname";

export function toPatient(pid: PID): Patient {
  const patient: Patient = { resourceType: "Patient" };

  patient.identifier = pid.$3_identifierList
    ?.map(toIdentifier).filter(Boolean);

  patient.name = pid.$5_patientName
    ?.map(toHumanName).filter(Boolean);

  if (pid.$7_dateOfBirth)
    patient.birthDate = toDate(pid.$7_dateOfBirth);

  if (pid.$8_administrativeSex)
    patient.gender = GENDER[pid.$8_administrativeSex];

  return patient;
}

05 · Connect

Connect any source

MLLP, HTTP and SFTP connectors bring in HL7v2, X12, C-CDA and FHIR — each written to a durable queue and handled the same way. Simulate a raw message to test a connector before it goes live.

MLLP, HTTP and SFTP connectors
HL7v2, X12, C-CDA and FHIR out of the box
Per-facility throughput and connection status

▾MLLPhospital-mllp

10.42.1.18:66612 fac

└Hospital EHR · Main38/m

└Hospital EHR · Lab6/m

▾MLLPclinic-mllp

10.42.3.7:66611 fac

└Clinic EHR · Ambulatory32/m

▾MLLPed-mllp

10.42.4.21:66621 fac

└ED system9/m

▾HTTPlab-fhir-http

planned connector—

▾SFTPregistry-x12-sftp

planned connector—

MLLP

hospital-mllponline

10.42.1.18:6661 · TLS

View messages Test

Protocol

MLLP · TLS

Host

10.42.1.18

Port

6661

Facilities

Simulate message

Paste a raw message here — HL7v2, X12, C-CDA or FHIR… MSH|^~\&|SENDER|FACILITY|…

Send

Throughput

Facilities2

Total messages1,820,442

Last minute44/m

Last hour2,110/h

06 · Reference

The specs, built in

Browse HL7v2 segments, fields and tables right inside the console — the same reference your mappers lean on and the AI assistant reads when it writes a converter.

Segments, data types and tables
Field cardinality, length and usage
The spec the AI assistant reads, one tool-call away

HL7v2 · v2.5.1 Copy link

Filter spec…

▸Messages8

▾Segments130

MSH

EVN

PID

PV1

OBX

NK1

OBR

IN1

DG1

▸Data types84

▸Tables412

PIDPatient IdentificationSegment

Fields10

Pos	Field	Type	Card	Len	Usage
PID-1	Set ID - PID	SI	0..1	4	O
PID-3	Patient Identifier List	CX	1..*	–	R
PID-5	Patient Name	XPN	1..*	–	R
PID-6	Mother's Maiden Name	XPN	0..*	–	O
PID-7	Date/Time of Birth	DTM	0..1	–	O
PID-8	Administrative Sex	CWE	0..1	–	O
PID-10	Race	CWE	0..*	–	O
PID-11	Patient Address	XAD	0..*	–	O
PID-13	Phone Number - Home	XTN	0..*	–	B
PID-14	Phone Number - Business	XTN	0..*	–	B

Components

Where Interbox sits

Interbox sits between the systems that send data and your FHIR server. A built-in source and sender bracket the mapper you own — and any message that can't proceed drops to the dead-letter queue below instead of being lost.

Upstream source systems

Hospital EHR

HL7v2 · MLLP

Clinics & ED

HL7v2 · MLLP

Reference lab

X12 · SFTP

Partner API

C-CDA / FHIR · HTTP

Interbox integration engine

Source

built-in · parses onto the queue

Queue

Postgres · one row + status per message

Mapper

your code · typed segments, concept maps

Sender

built-in · one FHIR Bundle, content-hashed

Downstream FHIR destination

Aidbox

native · content-hashed

Any FHIR server

transaction Bundle

Your apps, analytics and exchange then read standard FHIR.

Dead-letter queue

Every error and deferred message collects here, off the main flow — searchable, with its cause attached.

fix & reprocess → back to the queue

Message lifecycle

A status on every message

The happy path is three steps. A message that can't proceed is flagged error — and you can defer any error to decide on it later. Nothing is dropped.

received

Accepted and written to the durable queue.

processed

Converted to FHIR, ready to deliver.

sent

Delivered to the FHIR server.

error

Bad data the message can't recover from. Tagged with a kind — unmapped_code, validation_error, mapping_error — plus its cause, so you can classify and triage it.

→ fix and reprocess, or defer it

deferred

You move an error aside by hand, without resolving it — postponing the decision when the fix isn't obvious, like an upstream problem.

→ revisit whenever you're ready

Pricing

Simple, predictable licensing

Build for free in development. Go to production standalone, or at a lower rate alongside an Aidbox license.

Developer

Free for development

The full engine and SDK for building and testing — no PHI.

All built-in stages and your own mappers
Operations UI, AI assistant, hot-reload
Synthetic / non-PHI data only

Try now

With Aidbox

Best value

$1,200 / mo

or $12,000 / year

The same Interbox at a lower price when you buy it together with an Aidbox license.

Production use with PHI
Any FHIR server, native to Aidbox
Lower rate with an Aidbox license

Book a demo

Standalone

$1,900 / mo

or $19,000 / year

Run Interbox in production against any FHIR server.

Everything in Developer
Production use with PHI
Any FHIR transaction-bundle server

Book a demo

Engagement & Team

Want us to build it with you?

Our engineers work alongside yours — converters for your message types, EHR wiring, operations handoff. You ship faster; your team owns it after.

Custom scope — pricing on request.

Book a demo

FAQ

Questions, answered

Talk to us about your integrations

Tell us about your sources and your FHIR destination — we'll come back with a concrete shape for the work.

Address: Health Samurai Inc. 1891 N Gaffey St Ste O, San Pedro, CA 90731
Telephone: +1 (818) 731-1279
Email: hello@health-samurai.io