New updates and improvements to Orion.
2.1.0
2.0.2
2.0.0
The wait is over — literally. Orion now performs comprehensive data analysis directly in chat. Ask a question, and Orion queries your connected sources, extracts the data, and executes code in a live notebook panel alongside the chat. You see every step as it happens: each extraction, each cell, each chart. Redirect at any point, interrupt long-running operations, or dig deeper into a specific finding without leaving the thread.
This replaces the old model where comprehensive analysis meant kicking off a background job and waiting 20-30 minutes for results. Now the conversation is the workspace – with the same accuracy and reliability as before.
When you're ready to generate a deliverable, Orion has already been pre-building it in the background during the analysis — so your report, slides, or dashboard build even faster.
You can also export any notebook as a .py script or .ipynb Jupyter file, and download extracted datasets as CSV — so your work is always portable. If you have existing analyses from Deep Analysis, you can migrate any of them into a live notebook with a single click and pick up where the original left off.
Two new live data connectors are now available for enriching analysis: connect a Weather source (no API key required) or to our provided News API, and Orion will pull real-time weather conditions and news context directly into your notebook.
Track the KPIs that matter most by pinning them directly to your project. Create a metric tile by asking Orion in chat. Metrics support a range of formats: single numbers, currency values, percentages, and full data tables with rich chart visualizations including bar, pie, heatmap, scatter, stacked, and combo charts.
Pinned metrics appear at the top of your project and are always visible to Orion's context — so when you ask "what's my revenue this week?" Orion answers instantly without making an extra query. Each metric has a detail view showing historical values over time, a run schedule, visibility controls (private, shared, or published), and health status tracking.
When you complete an analysis, you can ask Orion to produce beautiful, interactive, and fully customizable dashboards – in addition to the existing report and slides options.
Any output — dashboard, slides, or report — can now be published to a custom share URL and sent to your team. Set a human-readable slug (e.g. /share/q4-marketing-summary), and anyone with project access can view the published artifact without needing the original conversation. When paired with Launchbooks, published outputs update automatically on the schedule you set.
Once you've iterated on an analysis and landed on a workflow you trust, save the notebook to re-run on a scheduled basis and deliver results in your preferred format — a weekly slide deck, a daily report to your inbox, or a dashboard that refreshes on demand.
Subscribe to any saved notebook to receive email notifications when new outputs are ready. This closes the loop between exploration and automation: start in chat, iterate until the analysis is right, then set it and forget it.
You can now connect GitHub, Confluence, and Notion directly to your Knowledge Base and import pages using a built-in file browser. Choose Read & Write to bring content in as a fully editable page, or Link & Sync to keep it automatically updated from the source. Synced pages display a sync badge in the editor and a manual sync button for pulling in changes on demand.
This means your technical docs, product specs, runbooks, and company wikis can flow directly into Orion's context without manual copy-paste — keeping the knowledge Orion draws on current with the source of truth.
Orion now fully supports dark mode across the entire product. Every surface — chat, the Analysis Notebook, data extraction cards, dashboards, slides, reports, the Knowledge Base editor, and all chart types — adapts to your system preference or a manual toggle in settings. Charts update reactively when you switch themes, so your visualizations always look right whether you're working in light or dark.
You can now upload JPEG and PNG images directly in chat alongside documents. Orion can view images and graphics and incorporate what it sees into the analysis context — useful for sharing screenshots, charts from external tools, design mockups, or any visual data you want Orion to reason about.
1.2.9
You can now opt-in to receive helpful browser notifications (with sound) when Orion finishes responding in chat, even if you've switched to a different tab.
Connect your dbt Cloud environment directly to your workspace by providing your account credentials and environment ID. Your dbt model metadata syncs automatically, giving you instant access to your production dbt definitions alongside your other data sources.
1.2.3
Create a shared Knowledge Base within Orion to centralize company context, documentation, and institutional knowledge. With a built-in Markdown editor, author and maintain Wiki-style pages that boost conversation and analysis quality.
As you work, Orion can suggest improvements, make direct edits to Wiki pages, and help evolve your Knowledge Base over time, turning it into a living source of truth that gets smarter with every conversation.
1.2.1
Deep Analysis tasks now properly preserve relative date ranges like "yesterday" and "last month". When you rerun or schedule these analyses, they automatically use current data instead of relying on static date ranges. This makes it even easier to set any analysis on a recurring schedule.
1.2.0
You can now connect Databricks as a data source in Orion—both for Deep Analysis tasks and quick queries in Chat. Combine data from Databricks with your other connected sources (like BigQuery, Snowflake, or PostgreSQL) to get answers that span your entire data stack.
Manage memories directly within your project settings with a dedicated Memories tab. Search memories by name or content, filter by user, and delete memories you no longer need.
Orion now searches memories formed within a project before constructing data source queries, applying previously learned information about field naming conventions and filter values. This reduces mistakes and ensures consistent query patterns across your analysis work.
When analyzing unexpected patterns in your data, Orion can now search news articles and historical weather data to identify external factors that may explain anomalies—such as severe weather events, labor strikes, supply chain disruptions, or regulatory changes that impacted your operations.
Contact sales to get News & Weather data access enabled for your team.
1.1.16
You can now export Orion’s generated slide decks as PDFs for fast sharing. The PDF format preserves fonts, colors, styling, and layout with a clean 16:9 aspect ratio, making it easy to share presentations across different platforms.
PowerPoint exports are now faster and more reliable, with better support for complex styling including custom fonts, gradients, and rich text formatting. All exported slides remain fully editable in PowerPoint so you can make adjustments after download.
1.1.15
During deep analysis setup, you can now see the memories that influence the analysis instructions and insight report formatting.
Each memory displays with a delete option, allowing you to remove context that no longer applies to your analysis. When you remove a memory, Orion automatically updates your analysis instructions to reflect the change.
A bug was identified that could cause SQL queries used in Deep Analysis to use unintended LIMIT clauses, ultimately restricting the data available for analysis.
This fix ensures unintentional LIMIT clauses are removed during Deep Analysis so that the full (applicable) dataset is loaded for accurate results.
When you delete a project, all associated memories and user notifications are now properly removed from the system. Similarly, when a user is removed from a project, their analysis notifications for that project are automatically disassociated, preventing outdated notifications from appearing.
1.1.10
Setting up comprehensive, high quality analyses is now faster and more intuitive with multiple-choice clarifying questions.
When Orion needs more information to scope your analysis, you'll see structured options (A/B/C/D style) that you can answer with a simple click, or provide your own custom response when needed. This streamlines the setup process and helps ensure your analysis is properly scoped from the start.
SQL query results now display visual indicators when they contain data-limiting constraints like LIMIT clauses, date ranges, or sampling.
These badges appear in the query header and expand to show detailed constraint information, preventing you from misinterpreting partial results as complete datasets. For example, you'll immediately see if results are limited to 30 rows or filtered to a specific date range, helping you draw accurate conclusions from your data when Orion provides quick answers.
SQL query validation and auto-fix functionality now works with Snowflake datasources, not just BigQuery. When a query returns zero rows, Orion can automatically suggest and apply fixes to help you get meaningful results.
1.1.9
Queries that return zero rows due to incorrect filter values now automatically recover. The system extracts the filter conditions from your query, looks up the actual values in your database, and intelligently rewrites the query with corrected values using fuzzy matching. This eliminates silent failures when filter values don't exactly match your data.
This release fixes a bug that occasionally removed mixed value columns in some report tables during the final validation and QA. Reports now load the correct tables even when your dataset contains tables with overlapping names or mixed value columns.
1.1.8
You can now connect multiple instances of the same datasource type or combine different data sources within a single project.
Configure separate production and development Snowflake environments, multiple Looker instances, or mix BigQuery with Snowflake to analyze data across your entire stack. Each instance gets a human-readable alias (like "Production Warehouse" or "Marketing Analytics") that appears in chat, making it clear which datasource you're working with at any time.
Orion will query and analyze across enabled data sources and surface any discrepancies or insights across your data infrastructure.
Generate presentation-ready slide decks directly from your insights. Simply ask Orion to create slides from an analysis, and it will produce an HTML slide deck with navigation controls and professional formatting. You can preview slides in-browser with fullscreen mode and export to PowerPoint format when ready to share.
Enterprise customers can now customize Orion with their company branding. Upload custom logos for light and dark modes, configure brand colors, and set company icons that appear throughout the interface and in PDF exports. Contact our team to enable white-label features for your organization.
1.1.7
Projects are now private by default and can be shared with specific team members. When creating or managing projects, you can invite users and assign them roles—Admins have full control, Analysts can create and edit analyses, and Viewers have read-only access.
This gives you granular control over who can see and work with your data while maintaining security and proper access boundaries.
Recurring analyses now use leverage Orion's internal context and memory to determine when to notify you. The system analyzes each insight and only sends emails when something meaningful has changed, automatically suppressing notifications when results remain stable.
You'll see warnings before Orion decides to limit certain notifications, and each notification includes a summary explaining what changed, reducing inbox clutter while ensuring you never miss important updates.
These settings are configurable from the analysis scheduling controls.
Create copies of existing projects with a single click from the project menu. You can optionally copy all members, files. and settings from the original project, making it easy to set up similar projects for different teams or use cases without manual configuration.
1.1.3
Your Recent Activity tab now prioritizes actively executing analyses, surfacing them to the top of your list for better visibility. Analyses are intelligently sorted by execution status first, then by most recent execution time, ensuring you always see what's currently running and what you've worked on most recently.
1.1.2
You can now control whether source citations appear in your insights at the project level. Access this new setting in your project's Insight Settings to toggle citation visibility on or off based on your team's preferences. This is particularly useful when sharing insights with stakeholders who prefer cleaner reports without technical attribution details.
Autonomous analyses (created and ran fully by Orion on it's own) now display a "Suggested" badge with an Orion avatar to distinguish them from manually created work.
1.1.1
Recurring analyses now use a more reliable scheduling system with more robust timezone support. This improvement addresses a bug where scheduled analyses would occasionally fail to run on the set schedule.
Enforce project-level data filters that automatically apply to all Looker queries (for Looker enabled projects only).
Admins can create rules like "always filter customer = Pepsi" to ensure that Orion only accesses authorized data without requiring users to specify filters manually.
1.0.4
Understand exactly where your insights come from with the new lineage viewer. Track the complete data flow from your original datasource through analysis steps to final reports, with inline SQL and Python code formatting for technical transparency. This makes it easier to verify analytical work and understand the transformation steps behind every claim.
You can now get in-depth explanations on exactly how Orion extracted, analyzed, and derived the data presented in the final insight. Simply highlight a portion of text and click the "Source" button to get an answer in the chat, including the queries and code snippets used.
Stay up to date with new features without leaving the product. Changelog notifications now appear directly in the interface when new updates are available, with persistent tracking so you won't see the same notification twice. Updates are managed centrally and display contextually based on where you are in the product.
0.0.3
Share your chat conversations with teammates using shareable links. When someone opens your shared conversation, they can create their own copy and continue the analysis in a new direction—perfect for collaborative exploration where different team members want to investigate different angles.
Stop in-progress responses at any time and immediately send a new message. No more waiting for lengthy responses to complete when you realize you need to clarify or redirect your question.
Orion now understands your local time zone and uses it to provide more accurate guidance on data availability and temporal context. When you ask about "today's data" or "this week's metrics," responses now reflect what those time periods are relative to your timezone.