We integrate with the stack you already run
Replacing a working system is expensive. The platform is designed to sit alongside the gateways, devices and records you already trust, and to keep them authoritative.
MikroTik RouterOS
Device API sync
FreeRADIUS
AAA and accounting
Cisco IOS
Config automation
Ubiquiti UniFi
Controller API
M-Pesa Daraja
STK push + callbacks
Stripe
Cards and subscriptions
PayPal
International checkout
HL7 v2
ADT, ORM, ORU messages
FHIR R4
Resource-level exchange
DICOM
Imaging study links
PostgreSQL
Primary data store
Redis
Cache and queues
SMTP / SMS
Notifications
Cloudinary
Asset delivery
QuickBooks export
Ledger export
Barcode (GS1)
Scan and print
Integration quality is an operational property
The difference between a working integration and a frustrating one is rarely the protocol. It is retry behaviour, idempotency, and what happens when something is missing.
Network
Devices are treated as an execution surface, not a data source. The platform pushes configuration and reads telemetry, so your routers hold the authoritative state of the link.
- Credentials scoped per device and rotated on a schedule
- Configuration applied as an idempotent operation
- Telemetry collected on a configurable interval
- Drift between intended and applied state is detected and reported
Payments
Payment integrations are built around reference-level matching. A transaction is only considered settled when it can be tied to a specific obligation, and every reconciliation exception is surfaced rather than swallowed.
- Outbound push to mobile-money and card gateways
- Inbound callback verification before any ledger posting
- Automatic matching against an invoice reference
- Unmatched transactions queued for human review
- Partial payments and refunds supported
Healthcare
Clinical interfaces follow published standards, but the hard parts are operational: identifier matching, result timing and deciding which system owns the record of truth.
- HL7 v2 messages for ADT, orders and results
- FHIR R4 resource-level exchange
- DICOM links for imaging studies
- Deterministic patient identifier matching with manual review for exceptions
- One authoritative source declared per data type
Data
The platform assumes PostgreSQL in production and Redis for cache and queues. Both are operated as boring, well-understood components you can inspect and back up yourself.
- Schema migrations versioned and reversible
- Row-level scoping applied to every query path
- Backup and restore rehearsed on a schedule
- Cache keys namespaced per tenant
Messaging
Notifications are triggered by workflow events, not by application code polling. Delivery status is recorded so a customer can always be told whether a message actually went out.
- Transactional email through your own SMTP provider
- SMS via a regional aggregator
- Per-tenant templates and sender identity
- Delivery receipts and bounce handling
Media
Assets — logos, product imagery, documents uploaded by customers — are stored on object storage behind a CDN, with tenant-scoped paths and signed access where required.
- Tenant-prefixed storage paths
- Image transformation and format negotiation
- Signed URLs for private documents
- Lifecycle rules for stale assets
Finance
The platform does not try to be your general ledger. It produces clean exports of what it actually knows about, so your finance system keeps its authority.
- Invoice, payment and adjustment exports as CSV
- Deterministic field mapping documentation
- Idempotent export jobs that can be re-run
- Reconciliation totals included with each export
Devices
Peripheral hardware is treated as a low-friction integration point. Scanners and printers should work on the first attempt without a support call.
- GS1 and supplier barcode formats
- Keyboard wedge and HID scanner support
- Thermal receipt printers over network or USB
- Label printing from product and batch records
Something not on this list?
Integration is delivered as a scoped service with a documented contract — message format, retry policy, reconciliation behaviour and a rollback plan.
- 1
Discovery
We confirm what each system owns, what it needs to exchange, and the expected volume.
- 2
Contract
A written data contract defines payloads, identifiers, error handling and reconciliation.
- 3
Implementation
Built with idempotency and explicit retry handling, then tested against the real counterpart.
- 4
Runbook
Operational documentation covering monitoring, replay and failure recovery.
Ask us about a specific integration
Tell us the systems on either side and what should flow between them. We will tell you whether it is supported, feasible or a bad idea.
Synthetic demo data only. No production credentials are ever exposed.