Book a Demo
Documentation · For IT teams

Appice IT Guide.

A reference for platform, infrastructure and integration teams evaluating or running Appice — what it plugs into, what it doesn't lock you into, and how it stays secure and available.

Integration No vendor lock-in Security Availability
01 · Integration

Plug into the systems you already run.

Appice is designed to slot alongside your existing core systems — not replace them. Real-time event streams and REST APIs land signals from any source; outbound connectors fire actions through whichever channels and providers you've already chosen.

Inbound: signals & sources

REST APIs, Kafka / Pulsar event streams, Webhooks, JDBC, and pre-built connectors for core banking, BSS/OSS, CRM, cloud data warehouses (Snowflake, BigQuery, Redshift) and lake formats (S3, ADLS, GCS, Iceberg).

Outbound: actions & channels

Push, in-app, SMS, email, WhatsApp, RCS, voice, branch and call-centre — through any provider. Bring your own SMTP, your own SMS gateway, your own WhatsApp BSP. Appice routes; you choose.

Identity & access

SAML 2.0, OIDC, SCIM provisioning, AD/LDAP federation, RBAC mapped to your IdP groups. Service-to-service auth via mTLS or signed JWTs.

Observability

Prometheus metrics, OpenTelemetry traces, structured JSON logs streamable to Splunk, ELK, Datadog or your in-house SIEM.

02 · No vendor lock-in

Your infrastructure. Your providers. Your call.

The decision system underneath your customer experience shouldn't be tied to a MarTech vendor's hyperscaler bet. Appice is deliberately built so the platform layers underneath are your choice — and remain swappable as that choice evolves.

Deployment

On-prem · Private · Public · Sovereign

Run Appice in your own data centre, your VPC, AWS, Azure, GCP, OCI, or in-country sovereign clouds. Same platform, same APIs, your residency rules.

Storage

Pluggable databases

PostgreSQL, MySQL, Oracle, MS SQL, Cassandra, MongoDB, Redis. Bring your own — Appice does not bind you to a managed database service.

Streaming

Bring your own bus

Kafka, Pulsar, RabbitMQ, Kinesis, Pub/Sub, Event Hubs. Swap message buses without re-platforming Appice.

Channels

Provider-agnostic

Switch SMS, push, email, WhatsApp, RCS or voice provider on a per-tenant basis without migrating data, models or audit history.

Models & analytics

Open formats

Models exportable in ONNX / PMML, decision logs in open formats, dashboards backed by SQL — not proprietary lock-in.

Containers

Kubernetes-native

Helm charts, OCI images, GitOps-friendly. Runs on EKS, AKS, GKE, OpenShift or vanilla k8s. Air-gapped installs supported.

See the full open architecture →
03 · Security

Security as architecture, not a checklist.

PII handling, network isolation, key management and audit logging are platform primitives — designed for environments where the regulator, the auditor and the CISO all need to look at the same answer.

Encryption everywhere

TLS 1.2+ in transit; AES-256 at rest. Bring your own KMS — AWS KMS, Azure Key Vault, GCP KMS, HashiCorp Vault, Thales / nCipher HSMs. Customer-managed keys supported end-to-end.

PII isolation

PII never enters the decisioning core. Inbound identifiers are tokenised at the edge into opaque, irreversible tokens; only those flow through models, journeys and logs.

Identity & access

SAML / OIDC SSO, SCIM, MFA enforcement, granular RBAC, just-in-time access for break-glass paths, full session and command audit trails.

Network & runtime

Private networking, VPC peering, PrivateLink / Private Endpoints, IP allow-listing, mTLS service-to-service, signed container images, CIS-hardened base images.

Audit & immutability

Every decision, override and admin action is logged with input, model version, output and operator identity. Logs are append-only and exportable to your SIEM.

Vulnerability & supply chain

SBOM published with every release. Continuous SAST/DAST, container scanning, dependency updates, third-party penetration testing, and a coordinated vulnerability disclosure programme.

04 · Availability

Built to stay up — under load and under failure.

Real-time decisioning has no batch window to hide in. Appice is engineered to keep the decision loop live during partial failures, peak loads and provider outages.

99.95%
Target platform availability (managed deployments)
<100ms
P99 decision latency at production load
RTO < 1h
Recovery-time objective (DR-enabled tiers)
RPO < 5m
Recovery-point objective with cross-AZ replication

Active-active topology

Decisioning, scoring and journey runtime stateless and horizontally scalable. Multi-AZ active-active by default, multi-region active-active for tier-1 deployments.

Graceful degradation

Channel-provider failover, model fallbacks, circuit breakers and bounded queues. A failing dependency degrades a single path, not the decision loop.

DR & backups

Encrypted backups, point-in-time restore, regular DR drills, runbooks reviewed with customer SREs. Restore tested — not just configured.

Capacity & load

Sized for tier-1 workloads — 10M to 200M+ customers, hundreds of thousands of decisions per second, with autoscaling tied to live signal volume.

Ready for an architecture review?

Bring your reference architecture, your compliance constraints and your hardest integration question. We'll map Appice into it on the call.

Book a Demo Risk Team Guide →