2026.09September 2026
Know your project at a glance
The catalog grows into a living health dashboard, compiled SQL previews render instantly in the browser, and lineage becomes something you can focus and share exactly as you see it.
Atlas CatalogCloud IDESlimmer CIEnvironments
New- Model health in the catalog: every model carries a healthy, caution, degraded, or unknown verdict with its last run status, plus performance history so you can catch a model trending slower before it becomes a problem.
- Local compiled SQL preview: see a model compiled instantly in your browser from your unsaved edits, with a one-click authoritative compile whenever you want the real thing. Secret-safe and adapter-aware.
- Focused lineage: anchor on a model and choose how many upstream and downstream hops to show, so tracing dependencies stays readable even in a large project.
- Ask Atlas in the catalog: open a node-grounded chat right where you are browsing, and document an undocumented model without leaving the catalog.
Improved- Richer catalog filtering: a labeled, collapsible filter panel narrows by type, schema, database, layer, health, and documentation or test status, with tests and semantic models now first-class, browsable entries.
- Hide-layer filters carry into share links, and you can multi-select specific models or hide a single model directly from the graph.
- Slimmer CI clarity: an explicit CI environment selector, a Compares against indicator, and each run showing the exact environment, warehouse, and user it ran with. Runs skip with a clear reason when no environment matches, instead of failing.
- Development environments can give each teammate their own schema, with the resolved schema shown right in the Variables tab so there is never any doubt where your models land.
- Workspace settings in one place: rename, pick an icon and accent color, and delete (owner only), with changes applied live across the editor.
Fixed- Commits now include brand-new files you just created, and discard works on deleted files so you can recover something removed by mistake.
- Share links resolve correctly for projects that live in a shared workspace, opening the exact focused and filtered view you intended.
2026.08August 2026
Built for the way you work
A focused push on the day-to-day: an AI co-pilot that can safely change code across files, a guided job wizard with a dry-run preview, and an in-editor Git and command experience that keeps you out of the terminal.
Atlas AICloud IDEVersion ControlSchedulingSlimmer CI
New- Atlas agent mode: ask for a change that spans several files and Atlas reads the relevant models, then proposes a complete, reviewable changeset grounded in your lineage. Nothing is applied until you approve each file.
- Atlas quick actions: paste an error to get a root-cause diagnosis and a minimal fix, or describe a business metric to draft its governed definition.
- Guided Create Job wizard: choose how a job triggers (scheduled, manual only, on merge, or both), preview the exact commands and selection in a server-side dry run, and let cycle-safe validation block invalid job chains.
- dbt command bar in the editor: type any dbt command with command and selector autocomplete, recall history with the arrow keys, and rerun any past command in one click.
- Save a worksheet straight into your project as a version-controlled model, with a folder picker.
Improved- Defer to production, favor-state, and a source freshness pre-check let jobs build only what changed while resolving everything else against trusted production data, with a readiness panel and an SLA-aware cadence hint.
- Smarter in-editor Git: a live branch-state strip (ahead, behind, one-click merge base), auto-drafted commit messages, a PR preview before you open it, a side-by-side merge editor, and inline line blame.
- Slimmer CI run management: a pull-request-first run list with status, PR, and date filters, archivable hidden runs, pagination, and deep links from PR comments straight to the run.
- Actionable CI failure reports show which nodes failed and why with a log tail, and hand off to Atlas for a grounded fix in a couple of clicks.
- Connection access levels keep production warehouses safe: share with the workspace, restrict to admins and owners, or keep private.
- Editor productivity: a command palette, project-wide find and replace, right-click file tree actions, and impact preview before a run.
Fixed- New branches are cut from the live head of the source branch, and a deleted branch falls back gracefully instead of stranding you.
- The DAG panel shows a clear notice for generated nodes with no source file and falls back to the server graph when the working tree cannot be read.
2026.07July 2026
A programmable platform
The public API grows into a full management surface, jobs become multi-step pipelines, and moving over from another platform is now a guided, reviewable process.
Public APISchedulingMigrationGovernance
New- Public API expansion: create, update, and delete scheduled jobs, read run logs and artifacts, and browse environments, connections, and the catalog, all through scoped tokens.
- Ordered job steps: run several dbt commands in sequence on a single job, stopping at the first failing step.
- Job chaining: a successful run can automatically enqueue another job in the same project, with cycle protection built in.
- Guided migration from dbt Cloud: upload an export, validate it, preview every action in a dry run, then apply the import idempotently.
- Managed dev databases: provision a governed, time-limited clone of a source database on demand, with automatic cleanup.
Improved- Catalog and CI now surface dbt governance metadata: contracts, access levels, groups, versions, and deprecation dates.
- Slimmer CI weighs contracted and public models as higher risk so review severity reflects real impact.
- Create a pull request in one click with a deep link to your provider, pre-filled with branch and base.
2026.06June 2026
Automation and access, on your terms
Service accounts bring first-class programmatic access, and workspace-aware databases keep large teams organized.
Service AccountsAPIWorkspaces
New- Service accounts: machine identities with scoped, least-privilege API tokens for CI pipelines and orchestrators.
- Per-token controls: expiry windows, request rate limits, and optional IP allowlists, with reveal-once secrets you can rotate or revoke.
- Interactive API reference: browse every supported endpoint with parameters, a ready-to-run example, and a sample response.
- Project discovery endpoint so automation can find its own way from projects to jobs to runs.
Improved- Workspace-scoped databases: filter connections by workspace and see at a glance which workspace each belongs to.
- Environment variable validation flags missing, unused, and mutually exclusive credentials before a run starts.
- Snowflake account entry now cleans up pasted host suffixes automatically.
- Set a default landing workspace so you start each session where you left off.
2026.05May 2026
Connect everything
A central place for reusable credentials, richer post-run automation, and safer sandboxes for every developer.
IntegrationsSchedulingEnvironments
New- Integrations: a central credential store for Power BI, API endpoints, and PagerDuty. Define once, reference from any job or alert.
- Post-run actions: refresh a Power BI semantic model or call an API endpoint automatically when a job finishes.
- Sandbox environments: per-developer schema isolation so feature-branch builds never collide.
- Extended attributes: per-environment connection overrides with an adapter-aware editor and identity-key guardrails.
Improved- Connection tags with environment awareness and namespaced labels, plus a production run guardrail.
- New warehouse adapters: Microsoft Fabric, Azure Synapse, and Fabric Lakehouse.
- Scheduling gains a runs-overview chart with build duration, success rate, outcomes, and volume.
2026.04April 2026
Understand your project
A server-indexed catalog, shareable documentation, and Git-native data products bring clarity to every model.
Atlas CatalogData ProductsSharing
New- Atlas Catalog: a searchable, server-indexed documentation browser with a zoomable lineage graph, coverage scoring, and column-level detail.
- Data products: a Git-native catalog with governance metadata, SLA tiers, ownership, and reference validation.
- Share links: tokenized, read-only URLs to share your catalog or metrics externally, no login required.
- Catalog annotations: attach object and column-level references such as links, contacts, and notes.
Improved- Shared workspaces: organize projects into private, team, or org-wide scopes and copy projects between them.
- MotherDuck joins the list of supported warehouses.
2026.03March 2026
From models to metrics
Define and validate business metrics in one place, backed by an AI assistant that understands your project.
Atlas MetricsAtlas AISlimmer CI
New- Atlas Metrics: a shared catalog of semantic metrics with live SQL preview, branch-level diffing, and import from existing definitions.
- Atlas AI Co-pilot: a warehouse-grounded assistant that understands your schema, models, and active context, ready to use out of the box.
Improved- Atlas AI answers now draw on your governed metric definitions for consistent results.
- Slimmer CI adds a metric gate that blocks breaking changes to published metrics.
2026.02February 2026
Ship with confidence
Organizations, multi-version dbt, and an intelligent CI pipeline that reviews every pull request.
OrganizationsSlimmer CIdbt versions
New- Organizations: a global switcher with tabbed settings for members, budgets, and notifications.
- Slimmer CI: targeted builds, schema diffs, sample data comparisons, and AI-powered review comments on every PR.
- Per-environment dbt variables with auto-injection into manual runs, scheduled jobs, and CI builds.
Improved- Choose your dbt version per project, run, or schedule.
- Breaking-change and logic-change detection with merge impact alerts and one-click revert.
2026.01January 2026
The platform foundation
A complete cloud development environment for dbt: edit, explore, test, and deploy across any warehouse.
Cloud IDEMulti-WarehouseScheduling
New- Cloud IDE with schema-aware autocomplete, split views, and integrated run output.
- DAG Explorer, Worksheets, Database Explorer, and Schema Diff for exploring and validating your project.
- Support for eleven warehouse adapters across four Git providers.
- Scheduling, notifications, data testing, and audit logging to run production pipelines end to end.