What We're Building

End result:

• Connect to any Postgres (local, RDS, Supabase, whatever)

• Query using natural language (no SQL if you don't want)

• Build actual dashboards with filters and drill-downs

• Share with your team (live, not PDFs gathering dust)

• Schedule reports (because your CEO likes emails)

Tech:

• PostgreSQL (you have this)

• SyneHQ (Docker or cloud)

• Coffee (optional but recommended)

Step 1: Connect Your Database (~2 minutes, maybe 3)

First, you need to actually talk to Postgres. We use SyneHQ for this.

Cloud Version (Easiest)

1. Hit up data.synehq.com

2. Sign up (free trial, no credit card BS)

3. "+ Add Connection" → PostgreSQL

4. Fill in your db details

Pro tip: Create a read-only user. Trust me on this.

-- On your Postgres server
CREATE USER analytics_readonly WITH PASSWORD 'something_secure';
GRANT CONNECT ON DATABASE production TO analytics_readonly;
GRANT USAGE ON SCHEMA public TO analytics_readonly;
GRANT SELECT ON ALL TABLES IN SCHEMA public TO

Why? Because one day someone (probably you at 2am) will accidentally run DELETE FROM users and you'll wish you'd set this up.

Private Database Behind a Firewall?

Use TCP Tunnels:

1. Settings → TCP Tunnels

2. SSH tunnel: ssh user@yourserver.com -L 5432:localhost:5432

3. Connect via localhost

No firewall rule changes. No security team meetings. Just works.

Step 2: Ask Your Data Stuff (The Fun Part)

Instead of writing SQL, just... ask.

Click "Talk to Kole" (that's the AI thing). Select your database. Type:

"Show me total revenue by month for 2024"

What happens:

• Kole reads your schema (tables, columns, types)

• Generates the SQL

• Runs it

• Shows you results + suggests a chart

Generated SQL looks like:

SELECT 
  DATE_TRUNC('month', created_at) AS month,
  SUM(amount) AS total_revenue
FROM orders
WHERE created_at >= '2024-01-01'
GROUP BY 1
ORDER BY 1

Click "show as line chart" → instant viz.

Why This Beats ChatGPT

ChatGPT doesn't know your schema. It guesses. Usually wrong.

Kole reads your actual database:

• Real table names

• Actual columns

• Foreign key relationships

• Sample data

So when you ask "revenue by product", it knows you have orders.product_id joining to products.name. No guessing.

Step 3: Build the Dashboard (~3 minutes)

You've got one chart. Let's make it useful.

Create Dashboard:

1. "Dashboards" in sidebar

2. "+ New Dashboard" → name it whatever

3. "+ Add Chart"

Add Charts (Using AI Because Why Not):

Chart 1: Revenue trend

• "Line chart showing monthly revenue for 2024"

• Done

Chart 2: Top products

• "Bar chart of top 10 products by revenue"

Auto-generates:

SELECT 
  p.name,
  SUM(o.amount) as revenue
FROM orders o
JOIN products p ON o.product_id = p.id
GROUP BY p.name
ORDER BY revenue DESC
LIMIT 10

Chart 3: Regional breakdown

• "Pie chart, revenue by region"

Chart 4: Daily active users

• "DAU for last 30 days"

Add Filters:
Click "Add Filter" → Date Range → defaults to "Last 30 days"

All charts update in real-time. Magic.

Step 4: Share It (~1 minute)

Go to Teams in sidebar → Invite via email → Share dashboard URL

Automated Reports

Because your CEO needs weekly emails:

1. Workflows in sidebar

2. "Create new workflow" → Ctrl+M for AI help

3. Set frequency (daily, weekly, monthly)

4. Add recipients

5. Pick format (PDF or live link)

Example: Weekly revenue report, every Monday 9am.

Step 5: The Cool Stuff

Cross-Database Queries (Tangent Lakes)

Got data in MySQL + Postgres + CSV?

Join across them:

-- Query MySQL users + Postgres orders + Parquet exports
SELECT 
  u.email,
  COUNT(o.id) as total_orders,
  SUM(e.revenue) as export_revenue
FROM mysql.users u
LEFT JOIN postgres.public.orders o ON u.id = o.user_id
LEFT JOIN parquet.main.monthly_exports e ON u.id = e.user_id
GROUP BY

No ETL. No data warehouse. Just works.

Browser-Local Notebooks

Quantum Lab runs Python in your browser. Data never leaves your machine. Compliance teams love this.

SQL cell:

SELECT * FROM orders WHERE created_at > NOW() - INTERVAL '30 days'

Python cell:

import pandas as pd
df['month'] = pd.to_datetime(df['created_at']).dt.month
monthly = df.groupby('month').agg({'amount': 'sum'})

All local. No server.

Real Example: E-Commerce Dashboard

You run an online store:

orders (id, user_id, product_id, amount, created_at, status)
products (id, name, category, price)
users (id, email, country, created_at)

Dashboard you can build in 5 minutes:

• Revenue metrics (this month vs last)

• Daily revenue trend (last 90 days)

• Top 20 products

• Geographic breakdown

• Customer cohorts

Kole generates all the queries. You tweak if needed.

What About Metabase?

I've used both. Honest take:

Metabase:

• ✅ Free (self-hosted)

• ✅ Decent query builder

• ❌ Still requires SQL knowledge

• ❌ No AI

• ❌ 4-6 hour setup

Superset:

• ✅ Very powerful

• ✅ Great viz

• ❌ Steep learning curve

• ❌ Overkill for simple stuff

SyneHQ:

• ✅ Natural language (non-tech users can query)

• ✅ AI knows your schema

• ✅ Cross-database queries

• ✅ 10-minute setup

• ❌ Newer (less mature)

Best for: Startups with mixed teams (engineers + PMs + analysts)

Pricing Real Talk (10-person team)

Common Questions

Can I use my existing Postgres user?

Technically yes. But please create a read-only one:

CREATE USER dashboard_readonly WITH PASSWORD 'strong_password';
GRANT CONNECT ON DATABASE your_db TO dashboard_readonly;
GRANT USAGE ON SCHEMA public TO dashboard_readonly;
GRANT SELECT ON ALL TABLES IN SCHEMA public TO

Database is on localhost?

Three options:

1. SSH tunnel manually

2. SyneHQ TCP Tunnels (built-in)

3. Self-host SyneHQ on same server (Contact our team)

Works with RDS/Supabase/Neon?

Yep. Any Postgres endpoint:

• AWS RDS

• Google Cloud SQL

• Supabase

• Neon

• Railway

• CockroachDB

• Whatever

Is my data secure?

Cloud: Queries run on our servers, encrypted in transit, SOC 2 in progress

Self-hosted: Everything on your infra, you control everything

Browser-local notebooks: Data NEVER leaves your browser (WebAssembly magic)

Troubleshooting

Connection Refused

Check:

1. Is Postgres running? pg_isready -h your-host -p 5432

2. Firewall blocking SyneHQ IP?

3. postgresql.conf: listen_addresses = '*'

4. pg_hba.conf: host all all 0.0.0.0/0 md5

Permission Denied

Grant SELECT:

GRANT SELECT ON ALL TABLES IN SCHEMA public TO

Slow Queries

Add indexes:

CREATE INDEX idx_orders_created_at ON orders(created_at);
CREATE INDEX idx_orders_user_id ON orders(user_id)

Or use query caching (Settings → Cache TTL → 5 min)

Next Steps

Day 1: Add more connections, build team dashboards, set up reports

Day 2: Try Tangent Lakes, build retention analysis, funnel tracking

Day 3: Automate with Workflows, set up alerts

Day 4: Migrate team from Google Sheets, set up API access

Try It

Cloud: data.synehq.com — Free trial

Self-Hosted: Contact hello@synehq.com

Docs: docs.synehq.com

Early adopter: $9.50/user (first 1000 customers)

Questions? Hit me up @synehq

What are you building? Drop a comment 👇

Appendix: Sample Queries

Sales Queries

-- Revenue by day (last 30 days)
SELECT 
  DATE(created_at) as day,
  SUM(amount) as revenue,
  COUNT(*) as orders
FROM orders
WHERE created_at > NOW() - INTERVAL '30 days'
GROUP BY day
ORDER BY day;

-- Top customers
SELECT 
  u.email,
  COUNT(o.id) as total_orders,
  SUM(o.amount) as lifetime_value
FROM users u
JOIN orders o ON u.id = o.user_id
GROUP BY u.email
ORDER BY lifetime_value DESC
LIMIT 50;

-- Product performance
SELECT 
  p.name,
  p.category,
  COUNT(o.id) as units_sold,
  SUM(o.amount) as revenue
FROM products p
JOIN orders o ON p.id = o.product_id
WHERE o.created_at > NOW() - INTERVAL '90 days'
GROUP BY p.name, p.category
ORDER BY revenue DESC

Growth Queries

-- New signups by week
SELECT 
  DATE_TRUNC('week', created_at) as week,
  COUNT(*) as new_users
FROM users
GROUP BY week
ORDER BY week;

-- Retention cohort
SELECT 
  DATE_TRUNC('month', u.created_at) as cohort,
  COUNT(DISTINCT CASE 
    WHEN o.created_at BETWEEN u.created_at 
    AND u.created_at + INTERVAL '30 days' 
    THEN u.id END) as month_0
FROM users u
LEFT JOIN orders o ON u.id = o.user_id
GROUP BY