Core surfaces
Lakehouse shortcutsReference ADLS Gen2, S3, GCS, Dataverse, or another lakehouse without copying data — OneLake shortcuts in the Files and Tables tabs.CSA Loom Learn library
Hands-on walkthroughs and one guide for every Azure / Fabric editor Loom ships. Each topic links to the CSA Loom docs first, with Microsoft Learn a click away. The same guidance powers the in-editor "Learn about this item" drawer.140topics96editor guides16tutorials128Loom guides live
Tutorials
End-to-end walkthroughs that take you from an empty workspace to a working scenario.16
Core surfaces
Data pipelines & Mapping Data FlowOrchestrate Copy, Notebook, and Dataflow activities on the visual canvas; transform code-free with Dataflow Gen2; validate, run, and schedule.Core surfaces
Notebooks (Spark)Interactive PySpark / Spark SQL notebooks over the lakehouse: attach a cluster, read and write Delta, use %%sql, run, and schedule from a pipeline.Core surfaces
Warehouse SQLManaged T-SQL over OneLake: create and load tables, cross-database joins, DirectLake for Power BI, and serverless cost-per-scan discipline.Core surfaces
KQL / Real-time intelligenceEnd-to-end real-time path: Eventstream → KQL database → queryset + dashboard → Activator alert, all on the real ADX / Kusto data plane.Core surfaces
Purview governance (classic Data Map)Register sources, run scans, auto-classify sensitive data, then discover via catalog search, asset detail, lineage, and the glossary.Core surfaces
Cosmos Data ExplorerBrowse databases and containers, create and edit JSON items, and run NoSQL queries (nested objects, arrays, JOIN) against the real Cosmos data plane.Core surfaces
Deployment & BYOPush-button provision or reuse existing Azure services (bring-your-own) per service via EXISTING_* env vars, with honest infra gates and no drift.End-to-end tutorials
Your first workspaceProvision a workspace, set roles, and orient yourself in the Loom console.End-to-end tutorials
First Lakehouse + Delta tablesLand raw files, load them to managed Delta tables, and query via the SQL endpoint.End-to-end tutorials
Direct Lake parityWire a semantic model in Direct Lake mode with the warm-cache materializer.End-to-end tutorials
Activator rules over an IoT streamFire Teams / email actions when a streaming threshold is breached.End-to-end tutorials
Data Agent over a LakehouseStand up a conversational Q&A agent grounded in your lakehouse + semantic models.End-to-end tutorials
Mirror Cosmos DB to a LakehouseNear-real-time replicate a Cosmos container into OneLake and query the mirror.End-to-end tutorials
Publish a marketplace data productBundle items into a governed, endorsed data product and publish it to the catalog.End-to-end tutorials
Forward-migrate a Lakehouse to FabricLift a Loom lakehouse into Microsoft Fabric without re-engineering pipelines.Use cases
Real-world reference scenarios — government, healthcare, gaming, cyber, API-first — built on CSA Loom (Azure-native, never Fabric).21
Government
DOJ Antitrust AnalyticsAntitrust compliance + investigation analytics — step-by-step domain build on Loom.Government
Government Data AnalyticsGeneral government analytics platform on Azure-native Loom services.Government
DOT Transportation AnalyticsDepartment of Transportation data analytics end to end.Real-Time
Real-Time Anomaly DetectionFraud + anomaly detection on streaming data (Event Hubs → ADX → Activator).Industry
Casino & Gaming AnalyticsPlayer-grain facts, real-time win/loss, high-roller Activator alerts.Cybersecurity
Federal Cybersecurity & Threat AnalyticsThreat detection + security analytics on Loom.Platform
Unified AnalyticsConsolidated analytics — the Fabric experience on Azure-native Loom.API-First
API-First Multi-Model AI EcosystemAI + data through an API-gateway architecture (APIM).API-First
Enterprise Asset Management via APIMAsset management exposed + governed through API Management.Editor guides
One guide per item type — every Azure / Fabric editor Loom builds, with full feature parity.96
Fabric Apps
Rayfin appRayfin is Microsoft's open-source Backend-as-a-Service for Fabric (Build 2026 preview). You define data models, auth, APIs, storage, and business logic in TypeScript with the @microsoft/rayfin-core decorators, then `npx rayfin up` deploys it to your Fabric workspace as a Rayfin item — data lands in OneLake (no ETL) and inherits your tenant's Entra identity, security, and governance. In Loom you author the backend spec here and Loom generates the SDK model + the exact CLI commands to deploy; the Rayfin CLI runs on your dev machine.Preview
Loom guide coming
Data Engineering
LakehouseA Lakehouse is the unified store for files and Delta tables in OneLake. In Loom it rides on ADLS Gen2 for storage with the Fabric/Synapse SQL analytics endpoint and Spark for query. Use it as the bronze/silver/gold landing zone for any analytics workload.
Data Engineering
NotebookA Notebook is interactive Spark/Python authoring with cells and outputs. In Loom it attaches to a Synapse Spark pool or a Databricks cluster and reads/writes Lakehouse Delta tables. Use it for data engineering and data science work that needs code.
Data Engineering
Spark job definitionA Spark job definition runs a compiled Spark application (JAR or .py) headlessly against your lakehouse — like a notebook but for production batch. In Loom it submits a Livy batch to the configured Synapse Spark pool.
Data Engineering
EnvironmentAn Environment is a reusable bundle of Spark settings and libraries that you attach to notebooks and Spark job definitions. In Loom the spec persists to Cosmos and Apply to pool PUTs it onto the Synapse Spark pool.
Data Factory
Data pipelineA Data pipeline is visual ETL/ELT orchestration — Copy, Lookup, ForEach, Notebook, Stored procedure, Web and more. In Loom it shares run history with notebooks and dataflows and is backed by the Fabric Data Factory runtime.
Data Factory
Dataflow Gen2Dataflow Gen2 is low-code Power Query data prep with visual and M-code authoring. In Loom you read from the supported connector set, transform in the Power Query editor, and write to a lakehouse, warehouse, or SQL DB.
Data Factory
Copy jobA Copy job is wizard-driven bulk ingestion from any supported connector — source to sink, no transforms. In Loom a run materializes a Synapse pipeline and triggers it, with run history sourced from queryPipelineRuns.
Data Factory
Mirrored databaseA Mirrored database is a near-real-time replica of an external source (Azure SQL, Snowflake, Cosmos, Databricks, Postgres) into OneLake. Queries hit the mirror, never the source. Use it to join external data with lakehouses without re-ingesting.Data Factory
Mirrored Databricks catalogA Mirrored Databricks catalog mounts a Databricks Unity Catalog as a read-only mirror in OneLake. In Loom you query the Delta tables from Fabric without re-ingesting. Use it to bring governed Databricks data into Fabric analytics.Loom guide coming
Data Factory
Mounted Data FactoryA Mounted Data Factory is a read-only attachment of an existing Azure Data Factory. In Loom the run history and monitoring surface natively so you can run ADF pipelines without migrating them. Use it to fold existing ADF investments into Loom.Loom guide coming
Data Factory
dbt jobA dbt job runs a dbt Core project (models, tests, docs) against your warehouse or lakehouse. In Loom a run materializes a Databricks Job with a dbt_task and triggers run-now; the runs list comes from the Databricks jobs/runs API.Data Factory
Apache Airflow jobAn Apache Airflow job runs DAGs synced from a Git repo on managed Airflow (preview). Use it when you need Airflow operators (Spark, dbt, Snowflake) beyond what ADF/Synapse pipelines cover. This is a preview surface.Preview
Loom guide coming
Data Warehouse
WarehouseA Warehouse is a lakehouse-native T-SQL warehouse with separated compute and storage. In Loom storage lives on OneLake as Parquet and compute auto-scales. Use it for full T-SQL DDL/DML and DirectLake-mode Power BI.Databases
SQL databaseIn CSA Loom the SQL database surface is backed by real Azure database services — Azure SQL Database, SQL Managed Instance, and Azure Database for PostgreSQL Flexible Server — not Fabric SQL. It lists existing deployments across the subscription via ARM, lets you connect to one, provision new ones (ARM PUT), run SQL over the live TDS path, browse the schema, and register the database as a governed OneLake/Purview catalog asset.Loom guide coming
Databases
PostgreSQL Flexible ServerAzure Database for PostgreSQL Flexible Server (Microsoft.DBforPostgreSQL/flexibleServers) is a fully-managed PostgreSQL service. In CSA Loom you list existing servers across the subscription, provision new ones via ARM PUT, manage databases + firewall rules, browse schema, and register the server as a OneLake/Purview catalog asset. In-database query execution is an honest infra-gate until the pg driver + LOOM_POSTGRES_QUERY_LIVE are wired.Loom guide coming
Real-Time Intelligence
EventhouseAn Eventhouse is a compute-plus-storage container for one or more KQL databases that share compute. In Loom it is wired against the shared Loom ADX cluster. Use it as the home for real-time analytics on streaming telemetry.
Real-Time Intelligence
KQL databaseA KQL database is a Kusto store for high-volume, low-latency analytics over time-series, telemetry, and logs, with OneLake availability. In Loom it runs on the shared Loom ADX cluster and is queried with KQL.
Real-Time Intelligence
KQL querysetA KQL queryset is a persisted set of KQL queries with charts and saved views — like a report for raw streaming data. In Loom it runs against the shared ADX cluster and feeds Real-Time dashboards.
Real-Time Intelligence
Real-Time dashboardA Real-Time dashboard is a tile grid powered by KQL queries with parameters and auto-refresh. In Loom tiles render from the shared ADX cluster. Use it to monitor live telemetry with drilldowns and time-pickers.
Real-Time Intelligence
EventstreamAn Eventstream is a code-free visual canvas to ingest, transform, and route real-time event streams. In Loom you wire source connectors (Event Hubs, IoT Hub, Kafka, Azure SQL CDC) through optional transforms to destinations; pipeline config persists to Cosmos.Real-Time Intelligence
Event schema setAn Event schema set is a schema registry (Avro/JSON Schema/Protobuf) shared across Eventstream sources, KQL ingestion, and downstream consumers powering DeltaFlow CDC. In Loom subjects and schemas persist to Cosmos and the eventstream runtime reads them to validate ingress payloads.Loom guide coming
Real-Time Intelligence
ActivatorAn Activator (Reflex) detects conditions on a stream or KQL query and fires actions — Teams, email, pipeline, notebook, or Power Automate. In Loom it watches a real-time source and triggers automation with no code.
Data Science
ML modelAn ML model is an MLflow-backed registered model with versions and a PREDICT endpoint. In Loom it is wired live to the AI Foundry hub (Microsoft.MachineLearningServices/workspaces) via the BFF. Use it to register and deploy trained models.
Data Science
ML experimentAn ML experiment tracks runs, parameters, metrics, and artifacts for a model family using MLflow. In Loom it is wired live to the AI Foundry hub via the BFF. Use it to compare hyperparameter sweeps and promote the winning run.
Fabric IQ
OntologyAn Ontology defines business entities, relationships, and condition-action rules (preview). In Loom it types entities and feeds the graph backend semantic layer. Use it to give connected data a shared vocabulary.Preview
Fabric IQ
Graph modelA Graph model is the schema definition for a property graph — node labels, edge types, allowed properties, indexes (preview). In Loom it feeds Cosmos Gremlin, Cypher-over-ADX, or GQL backends. Use it to design the shape before loading data.Preview
Fabric IQ
PlanA Plan is declarative state for a set of items and their dependencies (preview) — like Terraform for Loom items: diffable, reviewable, applyable. Use it to capture and version the intended shape of a workspace.Preview
Fabric IQ
MapA Map is a geospatial visualization layered over Lakehouse, KQL, and Ontology data (preview). In Loom it is a map artifact bound to a geo-dataset, embeddable in reports and dashboards.Preview
Fabric IQ
Data agentA Data agent is conversational Q&A grounded in your data sources and semantic model. In Loom it is built on a Foundry prompt-flow plus AI Search hybrid retrieval over your warehouse, lakehouse, and semantic models.
Fabric IQ
Operations agentAn Operations agent monitors real-time data and recommends actions via Activator and Power Automate (preview). In Loom it watches items and workspaces, flags drift, opens incidents in the audit log, and proposes remediations via the Cross-item Copilot.Preview
Power BI
Semantic modelA Semantic model holds the tables, relationships, measures, and roles backing Power BI reports. In Loom it is wired against live Power BI REST via the Console UAMI. Use it as the shared business layer for reports, dashboards, and scorecards.
Power BI
ReportA Report is an interactive Power BI report with pages, visuals, and filters bound to a semantic model. In Loom it is reframed around embed, refresh, and export against live Power BI REST via the Console UAMI.
Power BI
DashboardA Dashboard is a pinned-visual canvas surfacing tiles from multiple reports. In Loom it is wired against live Power BI REST via the Console UAMI. Use it to monitor KPIs at a glance across reports.
Power BI
Paginated reportA Paginated report is a pixel-perfect RDL report for printable, parameterized output (formerly SSRS) — invoices, financial statements, regulatory filings. In Loom it is wired against live Power BI REST via the Console UAMI.
Power BI
ScorecardA Scorecard is a KPI tree with targets and status (OKR-style). There is no Fabric REST API for scorecards today, so in Loom this is metadata-only — the editor persists the KPI hierarchy and discloses the API limitation honestly.APIs and functions
Data APIData API builder (DAB) generates secured REST and GraphQL endpoints over a relational or Cosmos source from a single dab-config.json. In Loom the editor introspects the database schema, maps tables/views/SPs to entities with per-role permissions, relationships, and policies, emits the canonical dab-config.json, and (when a DAB runtime Container App is deployed) tests the live REST + GraphQL endpoints and publishes through APIM.Loom guide coming
APIs and functions
API for GraphQLAn API for GraphQL exposes a single GraphQL endpoint over Warehouse, Lakehouse, SQL DB, or mirrored databases. In Loom it auto-generates CRUD plus custom resolvers. Use it to give app developers one typed endpoint over your data.
APIs and functions
User data functionA User data function is Python (or C#) server-side compute with bindings to Fabric items and external connections, callable from notebooks, pipelines, and Power BI. In Loom it runs serverless with per-call billing.
APIs and functions
Variable libraryA Variable library is a centralized name-to-value store with value sets per environment (dev/test/prod). In Loom it is workspace- or domain-scoped and used for pipeline, notebook, and SQL parameter substitution.
Synapse Analytics
Synapse dedicated SQL poolA Synapse dedicated SQL pool is a provisioned MPP T-SQL warehouse (formerly SQL DW). In Loom it is wired via ARM REST for pause/resume and TDS query on workspace.sql.azuresynapse.net through the Console MI. It auto-pauses to control cost.Synapse Analytics
Synapse notebookA Synapse notebook is the Spark authoring surface in Synapse Studio — multi-language cells (PySpark, Spark Scala, Spark SQL, SparkR) run interactively on a Synapse Big Data pool via Livy. In Loom it reads/writes the workspace notebook artifact over the Synapse dev plane and runs cells against a live Livy session through the Console MI.Loom guide coming
Synapse Analytics
Synapse serverless SQL poolA Synapse serverless SQL pool is a pay-per-query T-SQL endpoint over ADLS — OPENROWSET, external tables, ad-hoc analytics, no compute to provision. In Loom it queries workspace-ondemand.sql.azuresynapse.net via the Console MI over a private endpoint.
Synapse Analytics
Synapse Spark poolA Synapse Spark pool is Apache Spark compute for notebooks and Spark job definitions on Synapse-managed clusters. In Loom it auto-scales and auto-pauses, sized by node family. Use it as the compute behind data engineering notebooks.
Synapse Analytics
Synapse pipelineA Synapse pipeline is the Synapse Integrate canvas — ADF-shaped pipelines, dataflows, and triggers that run inside a Synapse workspace. In Loom it reuses Synapse-attached linked services and integration runtimes.
Azure Databricks
Databricks notebookA Databricks notebook runs PySpark/SQL/R/Scala cells with cluster attach, Unity Catalog governance, and Photon execution. In Loom it is wired against the Loom-deployed Databricks workspace via Container App MI and AAD bearer tokens.
Azure Databricks
Databricks jobA Databricks job is a multi-task workflow — notebooks, JARs, Python wheels, dbt, SQL — with dependencies and retry policies. In Loom it runs against the Loom-deployed Databricks workspace via the jobs API.
Azure Databricks
Databricks clusterA Databricks cluster is all-purpose or job Spark compute — node types, autoscale, init scripts, libraries. In Loom it is managed against the Loom-deployed Databricks workspace. Auto-terminate controls cost.
Azure Databricks
Databricks SQL warehouseA Databricks SQL warehouse is a serverless or classic SQL endpoint over Delta Lake with Unity Catalog and Photon. In Loom it lists real warehouses and runs statements via /api/2.0/sql against the Loom-deployed workspace.
Azure Data Factory
ADF pipelineAn ADF pipeline is a classic Azure Data Factory pipeline — 90+ activities, integration-runtime-aware, on-prem via Self-hosted IR. In Loom it sits alongside Synapse and Fabric pipelines and reuses ADF linked services and IRs.
Azure Data Factory
ADF datasetAn ADF dataset is a typed pointer over linked services — JSON, Parquet, Delimited, SQL, REST, and more. In Loom it defines the source/sink shape used by Copy Data and Mapping Data Flow activities.
Azure Data Factory
ADF triggerAn ADF trigger is a schedule, tumbling window, storage event, or custom event trigger that invokes a pipeline. In Loom you wire one or more pipelines per trigger to automate ADF runs.Streaming analytics
Stream Analytics jobA Stream Analytics job runs continuous SQL-style queries over real-time streams (Event Hubs, IoT Hub, Blob) writing to Blob, SQL, Power BI, Event Hub, ADX, or Cosmos. In Loom it is listed and managed via ARM through the Console UAMI; the query persists to ARM via the transformations endpoint.Loom guide coming
APIs and functions
APIM APIAn APIM API is a versioned API on Azure API Management that auto-imports OpenAPI/GraphQL/WSDL and ties Loom items as backends. In Loom it is wired live to the deployed APIM instance; Save issues a real PUT.
APIs and functions
APIM productAn APIM product bundles APIs into a subscribable offering with rate limits, quotas, terms, and a publisher-portal landing. In Loom it is wired live to the deployed APIM; Save issues a real PUT.
APIs and functions
APIM policyAn APIM policy is inbound/backend/outbound/on-error XML applied at a scope — JWT validation, rate-limit, cache, transform, mock. In Loom you load the policy XML for a scope, it validates well-formed XML client-side, and Save issues a real PUT.
APIs and functions
Data productA Data product is a data-mesh-aligned package — dataset plus semantic contract, an APIM API, an access policy, and an owner — listed in the marketplace. In Loom the Publish-to-APIM button POSTs a real product as an idempotent upsert.
Azure AI Foundry
AI Foundry hubAn AI Foundry hub is an Azure AI Foundry hub workspace (Microsoft.MachineLearningServices/workspaces kind=Hub) — connections, models, online endpoints, computes, datastores, and jobs. In Loom it is the shared parent for projects, prompt flows, and evaluations.
Azure AI Foundry
AI Foundry projectAn AI Foundry project is a child workspace under the Foundry hub. It inherits connections, models, and datastores and scopes prompt flows, evaluations, and data assets. In Loom it is wired to its BFF route and discloses 503/notDeployed honestly.
Azure AI Foundry
Prompt flowA Prompt flow is a LangChain-style graph of LLM and tool nodes. In Loom you author the YAML/JSON definition, run it with inputs, and view run history via the Foundry BFF route.
Azure AI Foundry
Foundry evaluationA Foundry evaluation runs quality/safety/accuracy evaluators against a dataset plus a model deployment, surfacing metric tables and pass/fail signals. In Loom it is wired to the Foundry BFF route.
Azure AI Foundry
Content SafetyContent Safety is Azure AI Content Safety — text and image moderation across hate/violence/sexual/self-harm with severity thresholds. In Loom you configure thresholds and wire it in front of any LLM call.
Azure AI Foundry
Foundry tracingFoundry tracing surfaces operation traces (Application Insights) for prompt flow runs, evaluator runs, and endpoint calls. In Loom you filter by operation and time window to drill from a failed run into the actual span.
Azure AI Foundry
AI Search indexAn AI Search index is an Azure AI Search index — fields, scoring profiles, vector and hybrid query — that backs RAG grounding for Foundry agents. In Loom it is wired to the Foundry BFF route.
Azure AI Foundry
Foundry computeFoundry compute manages AML compute instances and clusters — create, start, stop, scale, delete. In Loom it is used by prompt flows, evaluations, and training jobs; auto-shutdown reduces idle cost.
Azure AI Foundry
Foundry datasetA Foundry dataset is an AML data asset — URI file, URI folder, or MLTable — versioned and used by prompt flows, evaluations, and training runs. In Loom it is wired to the Foundry BFF route.
Copilot Studio
Copilot Studio agentA Copilot Studio agent is a conversational agent stored in Power Platform Dataverse — instructions, knowledge, topics, actions, channels. In Loom it is wired live to Power Platform (BAP) and Dataverse via the BFF; tenant-gate errors surface as a MessageBar.
Copilot Studio
Copilot knowledge sourceA Copilot knowledge source grounds an agent — URL, file, SharePoint site, or Dataverse table. In Loom you pick an agent, then list and add sources via the Dataverse-backed BFF.
Copilot Studio
Copilot topicA Copilot topic is a trigger-phrase-driven dialog flow authored in Copilot Studio YAML. In Loom you pick an agent, list topics, and edit trigger phrases plus the flow YAML via the BFF.
Copilot Studio
Copilot actionA Copilot action is a Power Automate flow, custom connector, or prebuilt action bound to a Copilot Studio agent. In Loom you pick an agent and manage its action list via the BFF.
Copilot Studio
Copilot channelA Copilot channel publishes an agent to Teams, web chat, Direct Line, Slack, or a custom channel. In Loom you pick an agent and publish-to-channel via the BFF.
Copilot Studio
Copilot analyticsCopilot analytics shows sessions, resolution rate, escalation rate, and CSAT for a Copilot Studio agent (last 30 days by default). In Loom you pick an agent and view KPI cards sourced via the BFF.
Copilot Studio
Copilot template libraryThe Copilot template library is a CSA-curated gallery of agent templates — data steward, contract analyzer, RFP responder, and more. In Loom templates are Cosmos-backed and Use template creates an agent in the selected environment.
Power Platform
Power Platform environmentA Power Platform environment is surfaced via the BAP admin API — SKU, region, Dataverse domain, security group, and DLP summary. In Loom it is read live via /api/powerplatform/environments. Each prod/dev/UAT gets its own environment.
Power Platform
Dataverse tableA Dataverse table is an EntityDefinition — schema, attributes, primary keys, custom vs system — sourced from the Dataverse Web API v9.2. In Loom you pick an environment first, which drives the Dataverse base URL, then browse tables.
Power Platform
Power AppA Power App is a canvas or model-driven app in an environment — owner, last modified, play link. In Loom it is sourced from the PowerApps admin API after you pick an environment.
Power Platform
Power Automate flowA Power Automate flow is a cloud flow — state, trigger, run history, and manual run. In Loom it is sourced from the Flow admin API; you can list flows, inspect runs, and trigger a manual run.
Power Platform
Power Pages siteA Power Pages site (mspp_website / adx_website) is a low-code public-facing website over Dataverse — domain, status, type. In Loom it is sourced from the Dataverse Web API.
Power Platform
AI Builder modelAn AI Builder model (msdyn_aimodel) is prediction, extraction, classification, or form-processing — with state and status from Dataverse. In Loom it is sourced from the Dataverse Web API after you pick an environment.
Azure SQL Database
Azure SQL serverAn Azure SQL server (Microsoft.Sql/servers) is the logical container for databases — server-level admin, firewall, AAD admin, and the database list. In Loom it is read via ARM REST through the azure-sql-client.
Azure SQL Database
Azure SQL databaseAn Azure SQL database is a fully-managed PaaS database. In Loom you get a per-database T-SQL editor (TDS + AAD), Fabric mirroring config, geo-replication, and vector index — wired via ARM and TDS through the azure-sql-client.
Azure SQL Database
SQL Managed InstanceAn Azure SQL Managed Instance (Microsoft.Sql/managedInstances) gives near-100% SQL Server compatibility for lift-and-shift. In Loom this surface lists instances and state; editor execution (TDS via private endpoint) is deferred to a later v3.x release.
Azure SQL Database
SQL Server 2025 vector indexA SQL Server 2025 vector index is the native VECTOR type and index — CREATE VECTOR INDEX, JSON_AGG, regex, similarity search — for RAG without a separate vector store. In Loom it probes the SQL Server 2025 features against the target database.
Azure Geoanalytics
Geo mapA Geo map composes an Azure Maps account, style, and tile layer. In Loom it lists Azure Maps accounts via ARM when available and falls back to OSM tiles with a MessageBar when no Maps account is deployed. Map config is saved to item state.
Azure Geoanalytics
Geo datasetA Geo dataset is a GeoJSON or Parquet+geometry dataset in ADLS Gen2. In Loom the geometry-column inspector runs a sample T-SQL OPENROWSET against Synapse Serverless via the existing query route so you can preview the data.
Azure Geoanalytics
Geo queryA Geo query is a spatial query against Synapse Serverless or Kusto — H3, S2, ST_DISTANCE, ST_WITHIN. In Loom a KQL-or-TSQL toggle pre-populates H3 and ST examples and submits to Kusto or Synapse Serverless.
Azure Geoanalytics
Geo pipelineA Geo pipeline is an ADF/Synapse pipeline specialized for geo enrichment — H3 index, reverse geocode, buffer. In Loom it is a Cosmos-backed pointer to an ADF pipeline with a geo-enrichment flag; ADF integration is deferred to v3.x.Databases
Azure Cosmos DB accountAn Azure Cosmos DB account (NoSQL / Core SQL API) is a globally-distributed, multi-model database. In Loom the editor is a live Data Explorer over the env-pinned account (LOOM_COSMOS_ACCOUNT) — databases → containers → stored procedures / triggers / UDFs — driven by the real ARM control plane (Microsoft.DocumentDB/databaseAccounts, api-version 2024-11-15). Create/delete databases and containers (with partition key + manual/autoscale RU/s) run real ARM PUT/DELETE calls.Loom guide coming
Azure Graph + Vector
Cosmos Gremlin graphA Cosmos Gremlin graph is Cosmos DB for Apache Gremlin — graph traversal over property graphs. In Loom queries run via /api/items/cosmos-gremlin-graph/[id]/query (the gremlin npm client with AAD or account-key auth); a 501 surfaces if the runtime isn't configured.
Azure Graph + Vector
Cypher graphA Cypher graph lets Neo4j-trained engineers use the openCypher dialect; in Loom it is translated to ADX make-graph/graph-match operators and dispatched via the KQL database query route — server-side, no Spark or Gremlin, millisecond-scale up to ~10M edges.
Azure Graph + Vector
GQL graphA GQL graph uses the ISO/IEC 39075:2024 standard graph query language — vendor-neutral pattern matching. In Loom it is dispatched to the graph backend of record (ADX graph operators via the KQL query route).
Azure Graph + Vector
Vector storeA Vector store is a backend-agnostic vector index — Cosmos vCore, AI Search, or PostgreSQL pgvector — for similarity search and RAG grounding. In Loom you pick a backend and define an index spec, which persists to item state; a live similarity test is deferred to v3.x.
CSA Data Products
Data product templateA Data product template is a CSA-curated push-button bundle — medallion lakehouse, IoT analytics, federated mesh, RAG agent, geospatial. In Loom Instantiate POSTs to /api/items/data-product-template/[slug]/instantiate to spawn the underlying items.
CSA Data Products
Data product instanceA Data product instance is an instantiated data product in a workspace — composed of underlying items (pipelines, lakehouses, indexes). In Loom it shows the spawned components and a status table; health is best-effort from child items' updatedAt.
AI & Agents
Cross-item CopilotThe Cross-item Copilot is a natural-language orchestrator across every wired Loom service — Synapse, Lakehouse, Databricks, APIM, ADX, ADF, Power BI, Fabric, Foundry (25+ tools). In Loom it streams from POST /api/copilot/orchestrate via SSE and calls the same BFF actions the UI calls, with a full audit log.Data Factory
Logic AppA Logic App is an Azure Logic Apps (Consumption) workflow defined in the Workflow Definition Language (WDL): a trigger (Request, Recurrence) followed by actions (HTTP, ApiConnection, Compose, ParseJson, Query, Select, If/Switch, Response). In Loom it opens fully built-out from the installed definition or the live Microsoft.Logic/workflows resource, and Run trigger fires a real manual run.Loom guide coming
Service guides
How the Loom engines work under the hood (Activator, Mirroring, Direct-Lake).3
Loom engines
Activator engineHow Loom evaluates Activator rules and dispatches actions without Fabric Reflex.Loom engines
Mirroring engineThe change-feed replicator that keeps OneLake mirrors in sync with their source.Loom engines
Direct-Lake shimThe warm-cache materializer that gives Power BI sub-second Direct Lake reads.Reference
Architecture, parity, and orientation docs for the platform as a whole.4
Concepts
What is CSA Loom?The one-page orientation: what Loom is, how it maps to Fabric + Azure, and why.Concepts
Reference architectureEnd-to-end architecture of the Loom platform and its Azure backing services.Concepts
Portal architectureWhere admins and users go — the console surfaces and how they fit together.Concepts
Parity matrixFeature-by-feature parity scorecard against Microsoft Fabric and Azure.Looking for the complete documentation? Open the CSA Loom docs site