Only through SQL. Looker supports MongoDB through the MongoSQL interface, not natively. Looker generates SQL and requires data reachable through a SQL interface. To use MongoDB data in Looker, you must first build a relational SQL schema over it (mapping nested fields and arrays to tables) with MongoDB's SQL Schema Builder, or ETL it into a SQL warehouse like BigQuery or Snowflake. Knowi connects to MongoDB natively using MongoDB's own query language.

Mostly. LookML is required for most data sources in Looker; self-service Explores skip it for BigQuery and Snowflake. Governed Explores and dashboards are built on a LookML model that defines the tables, joins, dimensions, and measures. This provides strong governance but creates a developer dependency. Business users cannot modify LookML models, and most new data sources or schema change requires LookML development. Knowi has no mandatory modeling language.

Gemini in Looker offers conversational analytics, a LookML coding assistant (Preview), and Python-based advanced analytics. It benefits from Looker's semantic layer for accuracy. It is in an extended promotional period, after which quota-based pricing begins. Knowi AI is the default model, so inference is not routed to a third-party LLM unless you enable one. It works on unmodeled data without a semantic layer prerequisite, and includes Document AI for querying unstructured files. The core trade-off is governance-enhanced cloud AI (Looker) versus privacy-first AI on heterogeneous data (Knowi).

Looker's negotiated pricing model and annual commitments create a high floor before warehouse and ETL costs. Looker is mainly priced for enterprise deployments, though a Standard edition targets teams under 50 users. Smaller teams often find better value in platforms with predictable, lower-commitment pricing.

Not in LookML. All LookML joins must reference tables within the same database connection. You cannot join a BigQuery table with a Snowflake table in one Looker Explore. The workarounds are consolidating data into a single warehouse or Merged Results, which join up to 5,000 rows per query by default in Looker memory. Data Studio (formerly Looker Studio, the separate, lighter Google BI product) supports data blending across up to five sources, but with significant limitations. Knowi can join data from any combination of its SQL, NoSQL, REST API and file sources in a single query.

Not necessarily. If your organization has invested significantly in LookML models, your data is centralized in a SQL warehouse, and your governance requirements align with Looker's semantic layer approach, that investment has ongoing value. Knowi is most compelling as a replacement or complement when you have NoSQL or API data sources that Looker cannot reach directly, when you need cross-database joins, when data privacy prevents using cloud AI, or when LookML development has become a bottleneck for business user self-service.