The AI analytics platform for your whole team

Give your team a way to ask questions and automate their regular reporting. Orion connects to your data and business context to deliver accurate answers in any surface.

Example: Orion flags on the Tenwick marketing dashboard that sourced pipeline fell to 81% of plan in October. Someone asks why, and Orion answers: Leads are up, but pipeline isn’t following. Since the LinkedIn forms launched, they average 479 leads a week, while other forms fell from 1,004 to 869. Those leads also wait a median 30.2 hours for a first touch.It then shows how the answer was verified: the sourced pipeline vs. plan metric from the dbt semantic layer, a query of weekly leads by form source, and the team's W-shaped attribution definition.

Example: someone asks Orion, “Every Monday at 8am, send the pipeline report. If we’re under 90% of plan, alert #revenue in Slack and open a review room with the sales leads.” Orion builds a Weekly Pipeline Review workflow: a Monday 8 AM schedule, a pipeline vs. plan analysis in Snowflake, a weekly report emailed to the sales team, and a decision that checks whether pipeline is below 90% of plan. If it is, Orion posts an alert to the #revenue Slack channel and opens a review room with three sales leads. A test run finds pipeline at 82% of plan and sends the Slack alert.

Example: the Weekly Pipeline Review workflow opens an Orion Room with Marcus, Jordan, and Alex after pipeline lands at 82% of plan. Jordan says Northwind and Altura pushed their close dates. Alex asks Orion whether the gap is all Enterprise and what is holding those deals up. Orion answers: Mostly. Enterprise is $620K of the $900K gap, from three deals that slipped. Northwind and Altura have both been in security review for over 20 days. Marcus decides: Let’s get our security team on both calls this week. Jordan, can you set them up? Jordan replies that both calls will be booked by Thursday.

Example: Jordan asks Orion which deals are at risk this week. Orion answers: Two deals are at risk this week. Northwind Health and Altura Logistics have both been in security review for over 20 days. Orion notes that the Pipeline Review Playbook does not define an at-risk deal and suggests adding: A deal is at risk when it has been in security review for more than 14 days. Jordan sends the suggestion for review. In Orion’s Context review queue, Alex, the page owner, approves the change, and Orion uses the rule in future answers.

A real Orion dashboard.Give it a try

You see the work.
Every step, every time.

Hover or tap any underlined part of the answer to see where it came from.

What was revenue last month?

was in September, down

What Orion checked for this answer

Definition
Gross revenue sums order value before returns and discounts, excluding canceled and test orders. Priya Shah owns the definition.
Query
A logged, read-only Snowflake query sums September 2026 orders from fct_orders.
Assumptions
September 1–30 is compared with August 1–31, in USD across all regions.
Access
Dana Reyes can see region, channel, and segment. Customer names and emails remain hidden.

Reliable answers from your data, in the AI tools your team uses

See MCP Docs

Integrate Orion with Claude, ChatGPT, or Gemini Enterprise to deliver accurate, traceable answers using your data team's definitions and access controls.

Example question in Claude: Why did repeat purchases fall?

Example answer: Repeat purchase rate was 29.3% in Q3, down 2.1 points from Q2. The largest change came from first-time customers in the West placing fewer second orders.

Orion used your team’s approved repeat purchase definition and checked which records you can access. The query and source tables are available with the analysis.

Claude
Write a message...

Example question in ChatGPT: Why did repeat purchases fall?

Example answer: Orion’s approved repeat purchase rate was 29.3% in Q3, down 2.1 points from Q2. The decline is concentrated among first-time customers in the West, who placed fewer second orders.

Orion used your approved repeat purchase definition and only the customer records you can access. You can review the calculation and source tables in Orion.

Work with ChatGPT
GPT-6.1 Sol Medium

Example question in Gemini Enterprise: Why did repeat purchases fall?

Example answer: The repeat purchase rate was 29.3% in Q3, down 2.1 points from Q2. Most of the drop came from first-time customers in the West making fewer second purchases.

Orion applied your data team’s repeat purchase definition and checked the records you’re allowed to see. The underlying query is available in Orion.

Gemini
Ask Gemini

Collaborate and move
faster with Group Chats

Start a Free Trial

Your team's AI work is scattered across private chats in Claude and ChatGPT, making collaboration slow. Orion Rooms unites your team and AI in one conversation to ask follow-ups and analyze together.

Weekly signup review

Maya started this room
Maya added Leo, Priya, and Sam

Orion · Mon 9:00 AM

Weekly marketing report

Scheduled

Signup conversion

↓ 1.0 pt
Last week · 4.0%This week · 3.0%

Ask Orion...

Listening

Give your reports a group chat

Eliminate the back and forth when something changes. Schedule a report, dashboard, or deck with a Room and the right people already invited.

Ask Orion as a team

Anyone can ask a follow-up, question a result, or request another breakdown. Everyone sees Orion’s work as it happens.

Mirror the group in Slack

Connect a Room to a Slack channel and pick up the conversation wherever you do your work.

More ways to put Orion to work

Build dashboards, prepare reports, and share the results wherever your team works.

Build dashboards, decks, apps, and more

Describe what you want to track. Orion builds a dashboard your team can filter, explore, and share.

A booking analytics dashboard and a Terwick slide deck

Ask from Slack

Ask Orion a question in a channel or direct message and get the analysis without leaving Slack.

#analytics
2 online
Today

Orion: Morning, Sarah! I reviewed the last 30 days of order data — 2,145 one-time customers had strong first orders but never came back. A reactivation campaign could recover ~$27K in annual revenue.

Sarah K.: Interesting — what made their first experience strong?

Orion: Fast delivery (under 30 min), zero refunds, and above-average order value. I put together a business case with the full customer profile and three recommended actions. [Attachment: Reactivation Campaign — High-Potential One-Time Customers — $27K annual opportunity from 2,145 customers. Includes customer profile, reactivation strategy, and projected ROI.]

Sarah K.: Love it. Can you loop in Enrique? He runs campaign ops.

Orion: Done — shared the business case with Enrique in #campaign-ops and suggested a sync. I'll keep tracking this cohort and flag if the window starts closing.

Message #analytics

Listen to the briefing

Catch up on the weekly numbers with an audio briefing made from your analysis.

0:19

Audio sample transcript

Morning, team. Quick heads-up before you cut the ad budget. Traffic held STEADY, but signups fell twenty-five percent last week. Most of the drop came from MOBILE visitors on the new signup page, starting just after Tuesday’s update.

Honestly, I'd check the mobile form before changing the campaign.

The full breakdown is in your weekly report.

Share through portals

Give colleagues, clients, or partners access to content, with the access controls you choose.

Share Portal

Evaluate and improve productivity

See what your team is asking and improve answer quality in a click.

Answer quality health score, signals, and topics

Put Orion inside your product

Let customers ask questions about their data inside your app, with your branding and access set for each account.

Orion embedded in a branded customer app, with account-scoped access and a self-serve question

Give AI the context behind your data

Your data team knows how revenue is counted, which tables to use, and which exceptions matter. Orion brings those definitions, documents, and rules into every analysis your team runs.

Use your existing definitions

Bring in the metrics and business logic your team already maintains, and keep it fully portable with no vendor lock-in.

Import

Let Orion find what's missing

As your team uses Orion, it identifies missing or conflicting context and brings you questions to resolve.

Recent conversations

Looking for context to improve

0 of 4 found
Illustrative context gaps: renewal treatment in qualified pipeline, refund exclusions in gross revenue, the active customer cutoff date, and conflicting customer churn reporting windows.

Control what changes

Keep ownership clear, lock important context, and review proposed edits before they change what Orion uses.

Noah Cook wants to editAnalysis Framework

"We're not using basic out-of-the-box software. We're pushing the envelope with the best and brightest in the AI analytics space."

Kevin Dugan

Kevin Dugan

VP of Analytics & Marketing Intelligence

DMI Partners

Analytics keeps moving.
Orion is built for the next step.

  1. Dashboards

    Show what happened. Someone still has to ask why.

  2. Self-service

    More people can ask. Definitions drift and answers disagree.

  3. Governed AI layer

    One layer sets the definitions, access and logic every AI follows.

  4. Decision OS

    The answer, the owner and the next step live in one place.

See how Orion fits your stack

Pick the tools your team already uses. We’ll show you how they fit together with Orion and tailor a demo to your setup.

Choose your tools

Select as many as you use today.

1 / 2

Data and commerce

Where the numbers live

Business context

Definitions and documents

Where your team works

Where answers should appear

Choose at least one tool to continue.

Bring the questions.
We'll run it in a live demo.

Frequently Asked Questions

Learn more about how Orion helps companies unlock more from their data.

Orion is an AI analyst built by Gravity. It connects to your data warehouse, learns your business context through a centralized Knowledge Base, and delivers analysis as stakeholder-ready reports, slide decks, and interactive dashboards. It can run on a schedule, track KPIs, and be deployed as a customer-facing embedded analyst inside your product.

BI tools require someone to build and maintain dashboards. When executives have new questions, they wait for an analyst to produce the answer. Orion answers questions directly in natural language, produces the same quality of output (reports, dashboards, slides), and does it in minutes, not days. It also learns your business context over time, which dashboards can't do.

Orion reads the same business logic your reports run on (LookML, dbt, and more), so its numbers reconcile with what you already trust. When something doesn't match, Orion shows why.

A chatbot can answer a one-off question if you feed it context. Orion grounds every answer in your codified business logic, so the whole company works from the same rules, not one-off guesses. It checks its math, shows its work, and doesn't wait to be asked.

Natively to BigQuery, Snowflake, Databricks, Looker, Postgres, dbt, and more. It syncs context from GitHub, Confluence, and Notion, and delivers to Slack, Google Chat, Teams, and email.

Yes. On the schedule you set: to Slack, to email, or as a CSV attachment. Daily at noon, Monday mornings, whatever your team needs.

Our team helps you set up Orion in a few days. Knowledge Base configuration can be done incrementally and Orion becomes more aware of changes in your business over time, orienting itself autonomously.

Yes. Orion's embedded deployment lets you offer a white-label AI analyst inside your product, powered by your customers' data and your business logic. Comprehensive role based access controls scope visibility and data access by team, project, and permission tier.

Orion is permission-aware, so people only see the data they're already allowed to see. Access is scoped by team, project, and permission tier, and Knowledge Base changes go through approval workflows before they affect Orion's reasoning. SOC 2 Type 2 compliant.

No. Orion handles the routine analysis that consumes your data team's time (recurring reports, ad-hoc questions, metric lookups) so they can focus on strategic work. Think of it as adding a tireless senior analyst to the team, not replacing the team.

Customer Success Story

"Orion's ability to transform 40 hours of manual analysis into just one hour is a game-changer."

Read the Case Study