Describe It. See It. Ship It. AI Dashboards Are Here.

TLDR: We shipped AI-generated dashboards for Supabase. Type what you want to see—"Show me signups, revenue, and top features this month"—and get a working dashboard with metrics, charts, tables, and filters in seconds. No SQL. No configuration. Your data stays in your Supabase project.

---

Building analytics has always felt backwards.

You know what you want to track. You can describe it in plain English. But between you and that dashboard sits a wall of SQL queries, chart configurations, data transformations, and layout decisions. So you either spin up a heavyweight analytics tool, wire up event tracking, or just... don't.

Today, that wall comes down.

What We Built

Dreambase Dashboards transform natural language into fully functional analytics interfaces. Describe what matters to your business, and the AI handles everything else:

• Generates optimized SQL for each visualization type
• Selects the right chart based on your data shape
• Configures metrics, tables, and filters automatically
• Lays out components in a responsive grid
• Caches results in your own Supabase storage

The whole thing takes seconds. Not hours. Not a sprint. Seconds.

How It Actually Works

Here's the flow:

1. You type a prompt: "Show me daily active users, revenue by plan, and our top 10 features"
2. AI generates queries: Separate, optimized SQL for summaries, timeseries, and detail views
3. Data gets cached: Results stored as compressed JSON in your Supabase bucket
4. Components auto-configure: Metrics, charts, tables—all wired up
5. Dashboard renders: Interactive ECharts with unified filtering

The win: you describe outcomes, not implementation. The AI figures out how to query, aggregate, and visualize.

Multi-Dataset Architecture

This is the part I'm most excited about.

Traditional dashboards run one query and try to power everything from it. That's why they're slow, bloated, or both.

We took a different approach. Every dashboard uses 2-4 purpose-built datasets:

Each query is optimized for its job. Filtering happens client-side across all datasets simultaneously. The result: dashboards that render fast and filter instantly.

10+ Chart Types, Zero Configuration

We're using ECharts under the hood, which means serious visualization power:

• Cartesian: Bar, Line, Area, Scatter
• Proportional: Pie, Donut, Funnel
• Multi-Dimensional: Radar, Parallel
• Relational: Sankey, Graph
• Hierarchical: Tree, Treemap, Sunburst

Here's the trick: chart types within the same family share data structures. That means you can switch from a bar chart to a line chart instantly—no regeneration needed. Cross-family switches require new queries, but the AI handles that too.

Chat Refinement

Dashboards are rarely right on the first try. That's fine.

Instead of digging through configuration panels, just tell the AI what to change:

• "Add a breakdown by region"
• "Change that bar chart to show percentages"
• "Filter to only paid users"
• "Add a table showing individual transactions"

The AI updates queries and components in place. Iteration happens at the speed of conversation.

Your Data Never Leaves Your Supabase

This matters. A lot.

Dreambase stores query definitions and component configurations. Your actual data stays in your Supabase project. We cache results in a dreambase bucket in your storage—compressed JSON that powers the visualizations.

No data pipelines. No external data warehouses. No wondering where your customer data ended up.

Real-Time Refresh

Dashboards stay current two ways:

• Manual refresh: Click a button, queries re-run, cache updates
• Scheduled refresh: Set a cron (default: 30 minutes) and dashboards stay fresh automatically

Both approaches use the same underlying system: queries execute against your Supabase via the Management API, results compress and upload to your storage, and the client fetches the updated cache.

Who This Is For

If you're building on Supabase and want analytics without the overhead:

• Founders tracking metrics from day one
• Product managers who need KPIs without engineering tickets
• Ops teams monitoring operational data in real-time
• Customer success watching account health
• Developers who'd rather ship features than build dashboards

You don't need a data team. You don't need to wire up events. You need your Supabase connection and a sentence describing what you want to see.

What's Next

This is the foundation. We're already working on:

• Scheduled reports via Slack and email
• Dashboard embedding for customer-facing analytics
• AI-generated insights surfaced alongside visualizations
• More chart types and visualization options

But we're shipping this now because it already solves real problems. Waiting for perfect means waiting forever.

Try It

Connect your Supabase project, describe what you want to track, and watch a dashboard appear.

Get started at dreambase.ai →

---

We built this because analytics shouldn't require a data engineering degree. You know your business. You can describe what matters. The AI can handle the rest.

That's the bet we're making.