Universal Format Processing
N+M intermediate representation converts business, office, database, geospatial, and media formats losslessly—Parquet read targets exceed 1GB/s per core; XLSX exceeds 100MB/s per core.
Foundational Core
polyMAX
One embedded engine. 500 connectors. Schema drift, silos, and pipeline rot die here.
polyMAX is the embedded universal translation layer that turns disconnected systems into one executable data fabric. It is not a hosted iPaaS bolt-on and not a catalog pretending to be runtime: it compiles into the host product, eliminates the sidecar latency tax, and owns format conversion, protocol translation, connector execution, ETL/ELT, streaming, and CDP behavior in-process.
At substrate depth, polyMAX runs a Connector Development Kit of composable primitives—Source, Destination, AuthProvider, SchemaEvolution, StateManager, RateLimiter—wired through a Protobuf Connector Protocol. The catalog targets 500+ connectors (350+ sources, 150+ destinations) with sync modes from full refresh through incremental dedup, log-based CDC, and history-mode SCD Type 2. Schema drift is detected in under 100ms; typing and deduping preserve destination fidelity through raw-to-final table pipelines.
v2 expands polyMAX from integration-heavy CDP into a full data fabric and data mesh platform. Named runtime node classes—processing, cache, ETL, streaming, metadata authority, policy governance, connector gateway—carry hard placement law across plain, Kubernetes, Istio mesh, BYOC/Crossplane, hybrid, air-gapped, and constrained-edge estates. Domain data products, federated governance, and metadata-plane authority sit on top of active execution truth, not a passive catalog.
Protocol translation covers REST, GraphQL, gRPC, WebSocket, SOAP, MQTT, AMQP, and Kafka with bidirectional adapters and Enterprise Integration Patterns for routing, enrichment, splitting, aggregation, and circuit breaking. Stream processing delivers windowing and exactly-once semantics; batch engines handle checkpointed migration at warehouse scale. OpenLineage-grade column lineage, data quality metrics, and OMRS-aligned governance zones enforce who may see, move, and mutate data across tenant scopes.
polyMAX never ships standalone. It embeds inside DXP, XRP, Lugh-E, MUKR, Prometheus, Hephaestus, and the rest of the stack—consuming CRTL runtime contracts, publishing operation events to timeline-x, delegating replay legality to KAOS, evidence to ARGUS, workflow choreography to WFX-b, set context to fullSET, advisor intelligence to ATHENA, and governance budgets to MOIRA. One engine. Every pipe. No excuses.
polyMAX is an embedded engine, not a product you deploy alone. It does not replace your host application's UX, billing, or customer portal—it powers the data plane inside products that already own those surfaces. It will not invent deterministic replay, WAL recovery, or PITR semantics on its own; that authority lives in KAOS. It will not forge durable evidence or staged-before-success audit proof without ARGUS. It will not orchestrate multi-year sagas, backfills, or remediation playbooks without WFX-b. It will not adjudicate governance breakers or budget overrides without MOIRA. It will not carry causal event-spine provenance without timeline-x. It will not reason about operator trade-offs or simulate policy outcomes without ATHENA. It will not project manifests, endpoint resolution, or database HA posture without CRTL. And it will not magically fix upstream APIs that rate-limit you into the ground—polyMAX removes its own overhead, not vendor caps. Buy polyMAX for universal connectivity and fabric-grade execution. Pair it with the other cores for production-grade truth.
See it ignite.
A demo of one core is a tech talk. A demo of all nine is a reckoning. Book the second one.
Get in touch