Four sectors, four deployments

Each sector below names the Microcelium components actually running in that deployment. No abstract capabilities; the tech doing the work.

Platform infrastructure for regulated growth

PSP / Bank webhook
External payment event
Provider Network
REST, AMQP, MQTT abstraction
Ingestion Engine
Validate, wrap in MDXF envelope
Message Bus
RabbitMQ, durable, UUID-tracked
Prism API + Glass UI
Ledger, dashboards, RBAC

Parallel consumers

Policy Agent
Policy enforcement, auto-remediation (roadmap)
Medusa
Metrics, Loki audit trail, alerts

Both subscribe to the bus alongside Prism. No bolt-on observability, no separate compliance pipeline.

The Problem

  • Integration plumbing consuming engineering time
  • Compliance infrastructure rebuilt from scratch
  • Monitoring bolted on as an afterthought
  • Architecture that breaks at 100,000 users

How Microcelium Solves This

Provider abstraction

Connect once. Microcelium handles payment processors, banks, APIs through the Provider Network.

Organisation as Code

Define compliance controls, access and org structure in YAML. Changes propagate everywhere.

Built-in compliance

ISO 27001-aligned controls. Audit trails, access controls, GDPR tooling. Compliance agents monitor continuously.

Medusa observability

Prometheus, Grafana, Loki, Tempo from day one. See your entire stack.

Edge to cloud, headless

Edge devices
Pi Pico, ESP32, LoRaWAN sensors
Air / Sylph
Device orchestration, MQTT bridge
Ingestion Engine
Normalise, wrap in MDXF
Message Bus
Same RabbitMQ as everything else
TimescaleDB
Time-series storage, analytics

Parallel consumers

Pandora
ML serving over fleet telemetry
Medusa
Fleet health, device alerting

Edge data uses the same pipeline as every other event. No parallel IoT stack to maintain.

The Problem

  • Device management complexity
  • Fragmented data pipelines
  • Real-time processing at scale
  • Predictive maintenance needs

How Microcelium Solves This

Air/Sylph Platform

Device orchestration for Raspberry Pi Pico, ESP32, LoRaWAN. MicroPython-based edge computing.

TimescaleDB

Purpose-built time-series database for sensor data, analytics and long-term storage.

Same pipeline

Edge data flows through the same Ingestion Engine and Message Bus as everything else.

ML serving

Pandora runs ML inference over fleet telemetry. The serving fabric ships today; failure-prediction and anomaly-detection models are built per engagement, tuned to each client's devices and operational patterns.

Multi-client management, automated

OaC YAML
One entry per client: access, SLAs, integrations
Cargo Server
Per-client micro-machine: VM + storage + VLAN + firewall, all configured as code, Docker Engine ready
micro-glue
Generates the per-client Docker Compose stack and lands it on the Cargo Server
Prism API + Glass UI
Per-client portal, RBAC, dashboards

Cross-client services

Service Engine
Lifecycle, scaling, health per stack
Medusa
Multi-tenant dashboards, SLA alerting

Adding a client is a YAML entry and a commit. Everything downstream regenerates.

The Problem

  • Inconsistent client environments
  • Manual onboarding
  • SLA monitoring gaps
  • Service delivery overhead

How Microcelium Solves This

Portal scaffolding

Prism API and Glass UI primitives. Ticketing, invoicing, self-service provisioning and compliance dashboards become configuration, not a rewrite.

OaC Per Client

Each client gets their own Organisation as Code config. Isolated, consistent, auditable.

Automated provisioning

New client environments spun up through micro-glue and Cargo Servers automatically.

SLA monitoring

Medusa observability with client-specific alerting, dashboards and reporting.

ISO 27001 continuous, not annual

System event
Access grant, config change, auth
Message Bus
UUID-stamped MDXF envelope
Policy Agent Roadmap
Evaluates event against OaC policy
Decision
Allow · auto-remediate · escalate
Loki audit trail
Every step stamped, searchable

Policy & targets

OaC policies
Source of truth for every rule
Authentik
Target of RBAC remediation

Every decision carries the triggering event UUID. Audits are a query, not a sprint.

The Problem

  • Annual audit scrambles
  • Access control sprawl
  • Manual evidence collection
  • Reactive compliance

How Microcelium Solves This

BCDR Agent (roadmap)

Continuous ISO 27001 control-map monitoring. Evidence generation on demand. Drift detection feeds the audit trail. Paired with the Policy Agent for live enforcement.

Organisation as Code

Single source of truth for access controls. Changes tracked with full audit trail.

Audit-Ready Logging

Structured logging across all services via Loki. Searchable, timestamped, immutable.

Authentik SSO

Centralised identity management. RBAC enforcement across all systems.

Honest about scope

Not for pre-product startups

If you're still finding product-market fit, you don't need platform infrastructure yet. Build your product first.

Not for mature platform teams

If you have 10+ platform engineers and a working internal platform, you probably don't need us. We're for teams that can't justify that investment.

Which scenario fits?

Every deployment is different. Let's discuss your specific architecture challenges.