Skip to main content
Ever wonder which of your marketing efforts actually drive revenue? Most companies can’t answer this because their customer data lives in silos: marketing contacts in HubSpot, product usage in one system, payments in Stripe. You know the revenue numbers, but you can’t see the complete customer journey. In the next few minutes, you’ll watch AstroBee automatically discover connections between your data sources, then ask business questions in plain English and get actionable insights.
Prerequisites: Before you begin, create an account and sign in.

​
Get started

After signing in, you’ll see the onboarding screen with three options. Click Use demo data to load a pre-configured dataset that mirrors real-world data challenges.
AstroBee onboarding with three options
The demo represents a SaaS company with the same challenge you face: data in three separate systems that can’t naturally connect:
  • HubSpot Contacts - Marketing leads, prospects, and customer information
  • Product Users - Signups, logins, and feature engagement data
  • Stripe Customers - Subscriptions, revenue, and payment status
The problem: Email formats don’t match, IDs are inconsistent, and there’s no reliable way to track a customer’s journey from first contact to payment. The solution: AstroBee’s AI-powered Pattern Extraction Agent will automatically discover these connections for you.

​
Generate your data layer

After selecting the demo data, AstroBee begins building your data layer, a map of how your business concepts connect to your raw data and to each other. This is what lets you ask questions in plain language without knowing where the data lives.
Data layer building modal
Building a data layer generally takes 10-45 minutes depending on data complexity. For the demo data, expect a maximum of 10 minutes. Feel free to leave and come back later. AstroBee will continue working in the background.

​
Pattern Extraction Agent

Here’s where the magic happens. AstroBee uses a specialized sub-agent called the Pattern Extraction Agent to automatically discover how your tables connect. Watch it analyzing your data in real-time:
The agent goes through multiple phases:
  1. Phase 0: Initialize Workspace - Sets up the analysis environment
  2. Phase 1: Source Discovery - Scans all source tables and columns
  3. Pattern Discovery - Finds potential matches across systems using email, phone, and ID matching
  4. Build Entities - Creates unified business concepts from raw tables

​
Discovered patterns and validation

As the agent works, you’ll see patterns discovered between your systems. The Discovered Joins tab shows the connections AstroBee found automatically:
AstroBee automatically found these cross-system connections:
  • Product Users ↔ Stripe Customers via email
  • HubSpot Contacts ↔ Product Users via email
  • Product Events ↔ Product Users via user ID
  • HubSpot Contacts ↔ Stripe Customers via phone
Each pattern is validated with SQL queries to ensure accuracy. With these connections, you can track customers from first marketing touch to revenue, see which features drive retention, and answer questions that span all three systems. The agent completes with validation:
  • Coverage Check - All 4 source tables are represented
  • Connectivity Check - No orphan entities; everything is connected
  • SUCCESS! - Integrated Data Layer Generated Successfully
What just happened? In a few minutes, AstroBee automatically solved a problem that typically requires weeks of data engineering: connecting disparate systems with inconsistent identifiers. The Pattern Extraction Agent found email matches, phone matches, and ID relationships across HubSpot, your product database, and Stripe, without you writing a single line of code.Behind the scenes, AstroBee applied entity resolution and normalization to match records even when formats differ (like [email protected] vs [email protected]).

​
Explore your data layer

Once building completes, you can explore the entities AstroBee created. Click any entity to see its data:
Each entity has tabs for:
  • Data - Preview actual rows (200 rows shown with CSV download option)
  • Description - AI-generated explanation of the entity
  • Properties & Relationships - Schema details and connections to other entities
  • SQL - The underlying query defining this entity

​
Edit your data layer

AstroBee’s automatic pattern discovery is just the starting point. You can enhance your data layer using natural language. Toggle to Edit mode by clicking the pencil icon, then describe what you want to create: 
Create a unified health metric combining product engagement, billing 
status, and recency. This would enable proactive retention strategies 
by identifying high-value customers showing early churn warning signs.
In this example, we ask for a “Customer Health Score.” AstroBee analyzes your data layer, runs SQL queries to understand available data, and creates a new entity combining engagement metrics from Product Users, billing data from Stripe, and contact info from HubSpot. Your new Customer Health Score entity now appears in the Tables list alongside Company, Customer, and Product Event.

​
Ask cross-system questions

Time to see what was impossible before. Previously, answering “How much revenue is at risk?” would require:
  1. Exporting data from all three systems
  2. Matching emails manually in spreadsheets
  3. Cross-referencing with payment status
  4. Hours of data wrangling and potential errors
Now? You’ll simply ask the question in plain English. Click New Chat to start asking questions that span all your connected systems:
Try this query that was impossible before the data layer connected everything: 
Visualize at-risk revenue. I want to see exactly how much monthly recurring 
revenue sits at each risk tier to highlight where intervention saves the most money.

​
Chat response

AstroBee analyzes the data, understands your question, and generates a response that includes not just the answer, but business insights and strategic recommendations. AstroBee responds:
I’ll help you visualize at-risk revenue by risk tier to identify where intervention would have the greatest financial impact.
Risk tier breakdown:
Risk TierMRR% of At-Risk RevenueCustomersAvg MRR/Customer
At Risk$41,80568.5%172$243
High Risk$10,30716.9%51$202
Critical$8,91514.6%650$14
Strategic recommendation:
Focus retention resources on the “At Risk” tier first - it represents nearly 70% of recoverable revenue with customers still showing engagement signals. The average customer value ($243 MRR) also makes personalized intervention cost-effective.
This is the “aha” moment. You asked one question, but AstroBee gave you:
  • The exact revenue numbers you requested
  • A breakdown by risk tier with customer counts
  • Strategic recommendations for where to focus resources
  • Analysis that required data from three completely separate systems
Before: This question would require manual data exports, spreadsheet matching, and hours of work. After: A single natural language question delivers actionable insights instantly.

​
View the answer report

Click the Analysis Result card to view the full results panel:
The Results tab shows the answer as interactive charts and tables. For this question, AstroBee generated a bar chart showing MRR by risk tier and an aggregated data table.
Results tab showing chart and table

​
Try more questions

Now explore more cross-system queries. Each conversation is saved in the sidebar: 
Show a bar chart of engagement (total product events in the last 30 days)
by customer industry
AstroBee responds with key insights:
  • SaaS industry demonstrates 25% higher engagement than second-place Media industry
  • Total of 1,849 product events were captured across 10 industries
  • There’s a significant drop-off after the top 9 industries
The Sources & logic tab shows exactly how AstroBee joined Product Events → Customers → Companies to answer this cross-system question. Try asking about marketing channel performance, customer lifetime value, or users with high engagement but payment failures. Each question leverages the unified data layer to deliver insights that previously required hours of manual work.

​
Next steps

Ready to use AstroBee with your own data? Choose your path:

Connect data sources

Connect warehouses like BigQuery or sources like HubSpot, PostHog, and Salesforce

Connect to MCP

Query your data layers directly from Claude Code, Claude Desktop, or VS Code
Need help? Contact us at [email protected] for setup assistance.