Symbiosis Data Platform is the trust layer under your data: catalog with search, lineage, ownership, and docs; data contracts and schema-change management; quality tests and anomaly detection; freshness SLAs; and stewardship workflows. It also includes a CDP for identity resolution, unified profiles, consent, and activation-ready segments, with bridges to warehouses, Analytics, Knowledge, and Agent Cloud. Catalog, quality, and identity resolution are what BI, AI, and activation stand on.
Dashboards nobody trusts are worse than no dashboards. They train the organization to ignore evidence. Trust is engineered with lineage, tests, and freshness contracts, then earned back one reliable report at a time.
A CDP unifies records about customers into activation-ready profiles. A catalog unifies knowledge about all data into an inventory. Marketing wants the first; everyone silently depends on the second. Most enterprises asking for a CDP are actually bleeding from catalog problems. We ship both as distinct machinery in one product so they compound instead of confuse.
Every AI ambition is a data-quality bet. Agents and analytics inherit whatever the pipelines feed them. Pointing identity resolution at ungoverned sources merges two customers into one. That is a lawsuit, not a rounding error.
Data catalog
Search, definitions, owners, tags, and documentation over tables, streams, APIs, and files - discovery that stays linked to stewardship.
End-to-end lineage
Column- and table-level lineage from source to KPI to agent answer - impact analysis before changes land.
Ownership & stewardship
Named owners, stewards, and review SLAs - 'who owns this truth?' is a product field, not a Slack mystery.
Data contracts
Producer/consumer contracts with schema evolution rules; breaks fail in CI before dashboards lie.
Quality test suite
Freshness, volume, nulls, uniqueness, referential integrity, custom assertions - on critical tables continuously.
Anomaly & freshness monitors
Detect silent drift and late pipelines; page owners with lineage context, not just 'red.'
Access governance
Request/approve flows, RLS hints, audit of who queried what - catalog never becomes a permissions bypass.
CDP identity resolution
Unify customer identities across systems with precision/recall you can measure on YOUR ugliest sources.
Unified profiles & consent
Activation-ready profiles with consent tracking - activation blocked when consent is absent.
Segments & activation bridges
Segments to marketing/CRM/ads with audit; profiles cataloged as governed assets themselves.
Warehouse / lakehouse bridge
Works with Snowflake, BigQuery, Redshift, Databricks, Postgres - catalogs the engine you already run.
Pipeline observability
ELT job health, retries, SLA clocks - paired with quality gates on outputs.
Analytics semantic feed
Certified tables/freshness feed Symbiosis Analytics metric definitions - one trust chain.
Knowledge corpus feed
Documented datasets and data dictionaries become Knowledge collections with ACL mirroring.
Agent data plane
Agent Cloud agents query only certified, permitted datasets - contracts bound to tool scopes.
PII / classification
Auto-classify sensitive fields; retention and masking policies; residency tags.
Incident & war-room
Data incidents with blast-radius lineage, consumer notify, and postmortems.
Scorecards & trust badges
Dataset health scores visible in catalog and BI - uncertified data looks uncertified.
Migration & coexistence
Import from existing catalogs/DQ tools; coexist while certifying critical paths.
Eval for CDP merge quality
Golden identity sets; precision/recall regressions block resolution model deploys.
CATALOG PLANE
Search, docs, owners, tags, access requests over the estate.
LINEAGE & IMPACT
Table/column lineage, blast radius, change impact analysis.
CONTRACTS & QUALITY
Producer/consumer contracts, tests, anomalies, freshness SLAs.
STEWARDSHIP OS
Ownership workflows, trust badges, incident war-rooms.
CDP PLANE
Identity resolution, profiles, consent, segments, activation bridges.
PIPELINE OBSERVE
ELT health, SLA clocks, output quality gates.
TRUST BRIDGES
Analytics, Knowledge, Agent Cloud, warehouse connectors.
PRIVACY & COMPLIANCE
PII classification, retention, residency, evidence packs.
Investment is bespoke to estate size, compliance needs, and deployment shape - book a demo and we'll map the right fit.
STARTER
- → Catalog for critical datasets
- → Basic lineage
- → Quality tests on top tables
- → Ownership fields
- → One warehouse connector
BUSINESS
- → Everything in Starter
- → Data contracts + CI gates
- → Freshness/anomaly monitors
- → Stewardship workflows
- → Analytics & Knowledge bridges
- → Access request flows
- → CDP starter (scoped sources)
ENTERPRISE
- → Everything in Business
- → Full CDP + consent/activation
- → CDP merge eval harness
- → PII/residency packs
- → Agent data plane bindings
- → Incident war-room + evidence
- → Private deployment options
Data Platform, answered
Catalog vs CDP - which first?
Catalog if your problem is 'we can't find/trust data'; CDP if it's 'we can't see one customer across systems' on otherwise-sane data. They compound - catalog the CDP's sources as you go. When in doubt, catalog first; it's the smaller regret.
Can one product do both?
We bundle them in Data Platform, but the functions stay distinct: identity resolution and lineage governance are different machinery. Evaluate each on its own evidence.
Do we replace our warehouse?
No - Data Platform catalogs, contracts, and qualifies data in/around the warehouse or lakehouse you run. Trust layer, not a new engine by default.
How does this feed Analytics and AI?
Certified, owned, fresh datasets feed the Analytics semantic layer and Knowledge corpora; Agent Cloud can be bound to only those datasets under policy.
What if consent is missing for activation?
Activation is blocked - the right answer involves 'blocked,' not 'best effort.' Consent is a first-class CDP control.
How do we prove CDP quality?
Identity resolution on YOUR ugliest sources with precision/recall - not adjectives. Eval harnesses gate model changes.
See it against your reality.
A 30-minute walkthrough with an engineer who builds it - your use case, not a script.