Data Agents That Actually Work.
Organize your company's data. Put agents to work.
MinusX: Agentic Data Platform
Contextual. Reliable. Proactive.
DATA ENGINEERING AGENT
Self-improving Data Models with Automated Testing
1. EVAL QUESTIONS
| Question | Domain |
|---|---|
| Top 3 cities by orders? | Sales |
| Revenue by product category | Finance |
| User churn rate this month | Product |
| Daily active users trend | Product |
| Average order value by region | Sales |
| Customer lifetime value | Marketing |
2. Gather ClarificationsWhat is “active user” defined as?
3. Create dbt Models4. EVALUATE PERFORMANCE
| Question | Status |
|---|---|
| Top 3 cities by orders? | ✓ |
| Revenue by product category | ✓ |
| User churn rate this month | ✗ |
| Daily active users trend | ✓ |
| Average order value by region | ✓ |
| Customer lifetime value | ✓ |
PERFORMANCE
DATA ENGINEERING AGENT
Self-improving Data Models with Automated Testing
1. EVAL QUESTIONS
| Question | Domain |
|---|---|
| Top 3 cities by orders? | Sales |
| Revenue by product category | Finance |
| User churn rate this month | Product |
| Daily active users trend | Product |
| Average order value by region | Sales |
| Customer lifetime value | Marketing |
2. Gather ClarificationsWhat is “active user” defined as?
3. Create dbt Models4. EVALUATE PERFORMANCE
| Question | Status |
|---|---|
| Top 3 cities by orders? | ✓ |
| Revenue by product category | ✓ |
| User churn rate this month | ✗ |
| Daily active users trend | ✓ |
| Average order value by region | ✓ |
| Customer lifetime value | ✓ |
PERFORMANCE
DATA ANALYST AGENT
Accurate answers via governed data models and team specific context
Ad-hoc Analysis / Modify DashboardsModify SQL QuerySTATE OF THE COMPANY DASHBOARD
What were the main causes for the order spike in ~Aug?
Based on 2025 data, I see that there was a big spike in orders around July-August. Let me look into the following factors:
- • products
- • channel and regions
- • conversion rates
PROACTIVE ANALYTICS AGENT
Always-on Agent, monitoring all important metrics & generating real-time reports
EMAIL REPORT
“Let me know if there are any significant changes in revenue”
Hi Team,
Key Findings:
- • Online orders down 22%
- • Product category B affected most
- • Geographic: West region -30%
Recommended Action: Review pricing change for West region
PPT PRESENTATION
“Create a Quarterly Performance Review. Add region breakdowns for new initiatives”
Q4 Performance Review
SLACK INTEGRATION
“Monitor all our marketing campaigns. Tell me if anything stands out.”
Conversion for campaign fb_cta_b is ~4% (monthly avg conversion is 1.9%).
Accuracy and Trust
Text-to-SQL is Solved.
But that was never the bottleneck.
Existing workflows introduce friction at every step - combining fragmented data sources, modeling business concepts, or ad-hoc analysis. Most Text-to-SQL solutions rely on explicit step-by-step instructions and a lot of hope and prayer.
Manual, Slow & Laborious
- Best practices are difficult and slow
- Hard to modify by non-technical users
WITH user_ltv AS (
SELECT user_id, SUM(CASE WHEN status = 'completed' THEN total_amount ELSE 0 END) AS lifetime_value, CASE WHEN SUM(CASE WHEN status = 'completed' THEN total_amount ELSE 0 END) < 100 THEN 'low' WHEN SUM(CASE WHEN status = 'completed' THEN total_amount ELSE 0 END) BETWEEN 100 AND 500 THEN 'medium' ELSE 'high' END AS ltv_bucket
------- +86 more lines -------Error-prone & Untrustworthy
- Humans are still the bottleneck
- Results often unreliable
WITH user_ltv AS (
SELECT user_id, SUM(CASE WHEN status = 'completed' THEN total_amount ELSE 0 END) AS lifetime_value, CASE WHEN SUM(CASE WHEN status = 'completed' THEN total_amount ELSE 0 END) < 100 THEN 'low' WHEN SUM(CASE WHEN status = 'completed' THEN total_amount ELSE 0 END) BETWEEN 100 AND 500 THEN 'medium' ELSE 'high' END AS ltv_bucket
------- +86 more lines -------Accurate & Trustworthy
- Reusable, across different surfaces
- 100% reliable, confidence scores
SELECT
f.ltv_bucket, d.country, d.signup_source, COUNT(DISTINCT f.event_id) AS total_orders, AVG(f.delivery_duration) AS avg_delivery_hours FROM fact_orders f JOIN dim_user d ON f.user_id = d.user_id
WHERE f.status = 'completed'
GROUP BY f.ltv_bucket, d.country, d.signup_sourceMinusX In The Wild
Trusted by Fast-Moving Data Teams
"MinusX has completely transformed analytics at Habuild. The difference in data visibility by leadership and product team is night and day. Analyzing the effect of any initiative used to take more than a week. With MinusX, every member is empowered to look at data themselves and we now ask more questions!"
Cashboard
"Our users love AI Chat, especially when MinusX creates GUI Metabase questions that they can follow along. I recently did ~6hrs of SQL work in less than an hour with MinusX. The integration is seamless, support from MinusX team has been fantastic and we're excited for more future collaborations!"
"I’ve been using GenAI in Hex as well, and honestly, yours works much better. Can't wait to supercharge it with more context."
"I genuinely wouldn't have tried an analysis had I not had this tool."
"MinusX has effectively replaced the need for a dedicated data analyst."
