Access control
Schema- and field-level read/write/query rules by role.
API-first governance for teams that need more than open JSON storage—without standing up a separate MDM stack.
supplier_evidence
Tenant-scoped · schema-on-write · lineage on
01 Ingest02 Merge schema03 Unified records
source_data · auto_create_schema
Excel upload
{ "name": "Acme Parts", "tax_id": "FI-12345" }
WebApp form
{ "company": "Nordic Supply", "vat_number": "SE-98765" }
REST API
{ "supplier": "Baltic Co", "country": "EE", "contact_email": "ops@baltic.ee" }
Simulated flow — ingest heterogeneous payloads, merge schema on write, query unified records
Built-in data layer
“Operational data for workflows—not another database project.”
Platform
Datapool — governed data layer
Tenant-scoped schemas and JSON records wired to Process Automation, the REST API, and policy-bound agents. No extra database to provision for typical workflow data.
Datapool is a tenant-scoped operational data layer built into Tealfabric—not a separate database product.
Teams need structured records for subscribers, supplier evidence, process outputs, and agent context—but standing up a warehouse or MDM project slows down the workflows that actually ship value.
Datapools give you schemas, JSON document rows, merge-on-write ingestion, partial updates, and governance hooks—already wired to ProcessFlow, the API, and MCP read access for agents.
- Built-in data layer for automations—schemas, records, queries, and governance inside the platform
- Wired to ProcessFlow, the REST API, and policy-bound AI agents—no extra DB to provision
- Schemas that grow with your data via auto-create and auto-merge on write
- Safe multi-tenant isolation, audit, access rules, and quality scoring built in
- Agents read operational data responsibly through MCP tools with validated queries
- Virtual schemas connect external systems without copying everything first
Lead use cases
Operational data patterns teams ship first.
From supplier intake to agent-assisted queries—Datapools sit at the center of governed workflows, not beside them.
file → storage
row → source_data
auto_merge_schemaSupplier / form intake
File to tenant storage + structured row in Datapool with auto-schema merge.
PATCH subscriber
verify_token updatedSubscriber / CRM-lite lists
Partial updates to existing rows from process steps—patch semantics by default.
step result → insert
downstream queryProcess output persistence
Store step results for downstream steps, reporting, or agent context.
MCP datapool_query
SELECT … LIMIT 50Agent-assisted ops
“Show active customers from customers schema” via read-only MCP datapool_query.
virtual schema
cache TTL 300sExternal DB federation
Virtual schema over a connector-backed table with cached queries.
Schema-on-write in action
Heterogeneous payloads become one governed schema.
Excel uploads, WebApp forms, and REST payloads arrive with different field names. Datapool auto-merges new fields into the schema on write—then processes, APIs, and agents query unified records without upfront data modeling for every source.
supplier_evidence
Tenant-scoped · schema-on-write · lineage on
01 Ingest02 Merge schema03 Unified records
source_data · auto_create_schema
Excel upload
{ "name": "Acme Parts", "tax_id": "FI-12345" }
WebApp form
{ "company": "Nordic Supply", "vat_number": "SE-98765" }
REST API
{ "supplier": "Baltic Co", "country": "EE", "contact_email": "ops@baltic.ee" }
Simulated flow — ingest heterogeneous payloads, merge schema on write, query unified records
Core capabilities
Ship-ready operational data—schemas, ingestion, API, and process integration.
Tier A features are production-ready today: tenant schemas, schema-on-write, ProcessFlow access, partial updates, scoped REST API, and multi-tenant isolation.
/datapool UI
schemas · browse · paginateTenant-scoped schemas and records
Named schemas with typed fields, JSON rows, and tenant isolation on every operation.
Create and manage schemas in the tenant app—browse data in a spreadsheet-style table with sort, search, and pagination. Every row carries data_id, timestamps, and automatic tenant scoping.
auto_create_schema
auto_merge_schemaSchema-on-write ingestion
auto_create_schema and auto_merge_schema as payloads evolve.
First insert can create the schema; new fields from supplier submissions, WebApp uploads, or integrations extend the definition automatically. Schema version snapshots track definition changes.
$datapool.insert()
datapool.query()ProcessFlow-native access
$datapool in PHP steps and datapool in JS/TS sandbox—same REST surface.
Insert, update, query, and fluent builders inside process steps. Production pattern: store uploads and append rows with auto-schema merge into a source_data pool.
PATCH row
shallow mergePartial row updates
Shallow patch by default; replace_mode for full document replace.
Updates merge into existing data_content—ideal for “update verification token on subscriber row” workflows. Same semantics across Nest API, PHP legacy, and process sandbox.
/api/v1/datapool/*
datapool.read · writeREST API with scoped access
Full CRUD under /api/v1/datapool/* with datapool.read / datapool.write scopes.
Schemas, data, cache, virtual, and governance endpoints. Bulk insert supported—wire external systems and custom apps without a separate database project.
tenant_id scoped
every operationMulti-tenant by design
Every query and write is tenant-scoped automatically.
Isolation is built in—not something you wire yourself on a bring-your-own database. Promotable against stacks where each team provisions and secures their own store.
flow editor
dependency trackingProcess designer integration
Flow editor detects datapool usage in step snippets.
Query, insert, and update calls in step code are tracked for dependency and resource visibility—see what your processes touch before you ship.
Governance & power features
Governance without a separate MDM or data-quality stack.
Access control, audit logging, quality scoring, lineage, virtual schemas, and entity resolution—API-first for teams that outgrow simple JSON storage. Main tenant UI today focuses on schemas and data browser; advanced governance is primarily API and admin tooling.
Audit logging
Trail of schema, data, and query operations.
Operational changes are traceable for compliance and post-incident review.
Data quality
Validation rules and quality scores on insert/update.
Score incoming records at write time—useful for supplier evidence and intake programs.
Lineage
Source and transformation metadata on writes.
Graph and path APIs record where data came from and how it changed—agents and processes share the same lineage story.
Virtual schemas
Federate external integration tables with query cache and TTL.
Connect external systems without copying everything first—read-through and write-through cache for virtual queries.
Canonical / entity resolution
Match and link records by key fields.
Manual canonical links and entity resolution for teams outgrowing simple JSON document storage.
Agents & MCP
Agents can read your business data responsibly.
Policy-bound agents discover schemas via the datapool_schemas MCP resource, then run read-only datapool_query against tenant data—SQL-style SELECT with filters, ORDER BY, and LIMIT. Safer positioning than “full ANSI SQL warehouse”: the core engine is intentionally focused; advanced aggregates and JOINs are on the enhancement backlog.
datapool_query
Read-only SELECT against tenant schemas—agents query operational records in chat and automation context without copying data into a sidecar.
SELECT supplier_name, country FROM supplier_evidence WHERE status = 'active' LIMIT 50
datapool_schemas
MCP resource exposing schema metadata so agents know field names, types, and structure before they query—discover first, read second.
schemas: supplier_evidence, customer_sync_queue, subscribers
Who it is for
BizOps, data owners, and integration leads who need structured operational records without standing up a warehouse or MDM project first—supplier intake, subscriber lists, process outputs, and agent context on one tenant.
Related workflow: Customer & vendor intake. Platform depth: Process Automation, AI agents, Integrations, Governance.
Datapools are operational JSON storage on the tenant—not positioned as a petabyte data lake or full ANSI SQL analytics warehouse. Cross-tenant Datapool Library and object schema type are on the roadmap.
Operational data, not another database project
See Datapools inside Process Automation and agent workflows.
We walk through schema-on-write ingestion, partial updates, tenant isolation, and MCP read access—on the same runtime as your integrations and processes.