“What gaps did Sundial leave open in AI-native analytics, such as open semantic layers, real-time analysis, warehouse independence, or customer segments it did not serve?”
This report draws on 71 sourced rows scanned on 2026-10-24: 31 pages of Sundial's own documentation, which are treated as authoritative for Sundial's scope, and 40 market-comparator rows from 37 other hosts showing how other products occupy each opening. Six of the Sundial rows describe archived “classic” connectors, not current Context Engine support.
Definitions live as YAML in the customer's Git repo, but the format is Sundial-specific; MetricFlow import is a one-time bootstrap; direct non-MCP API access is limited and on request. Opening: an Apache-licensed, vendor-neutral semantic layer interoperable with dbt/MetricFlow, Cube, BI tools and agents.
Sundial documents always-live apps, refresh-triggered and scheduled tasks on a five-minute grid, and materialization schedules — but no Kafka/CDC ingestion, continuous streaming SQL, or subsecond operational alerting. Opening: continuous analysis and agentic anomaly response over event streams.
The Context Engine supports Snowflake and BigQuery; Redshift, Delta Sharing, JDBC, S3, GCS and Stripe appear only in archived classic docs. Opening: first-class Databricks, Postgres, ClickHouse, Trino, SaaS apps, files, and federated multi-source analysis without a warehouse migration.
The terms frame use as the customer's internal business purpose; MCP is the preferred programmatic surface; direct API access requires a Sundial representative. No white-label, multi-tenant embedding SDK is documented. Opening: AI-native embedded analytics for SaaS product teams.
Despite every-decision-maker positioning, successful use assumes an existing warehouse, governed metrics, admin/data-team ownership and often dbt/Git workflows, with Sundial's Applied AI team setting up context in the first 30 days. Opening: self-serve analytics for startups and operational teams on SaaS tools.
Notion, Slack, Linear and GitHub context connectors are explicitly read-only, and scheduled analyses deliver into Slack only. No governed write-back into CRM, ticketing or finance systems is documented. Opening: analysis-to-action agents with approvals, audit trails and reversible write-back.
| Provider | Type | Capability or positioning | Docs | Source |
|---|
Method: 71 sourced rows scanned 2026-10-24 — 31 Sundial documentation pages (6 of them archived classic-connector pages, marked “archived”) and 40 market-comparator rows across 38 independent hosts, one row per URL, no URL above 1.5% of rows. The matrix columns are the six openings named in the brief; each dot is one comparator assigned editorially to the openings its sourced positioning addresses, as directional evidence, not a scored ranking. Capability text is truncated for space; full detail sits behind each row's link.