The next step in Headless Analytics is here

Headless Analytics is a simple idea: your data should be ready to work wherever work is happening.

Not trapped behind a dashboard. Not buried in a production database. Not rediscovered from scratch every time someone opens a new agent session.

It should show up, correctly and safely, when your team needs an answer in Claude Cowork, Claude Code, ChatGPT, Codex, Cursor, an internal workflow, or whatever agent surface comes next.

Today, we are opening early access to the Dreambase MCP.

This is the next major step in the Headless Analytics journey at Dreambase. It turns the analytical foundation we have been building inside Dreambase into tools that your agents can use directly: blazing-fast access to pre-calculated business metrics, validated historical datasets, and the skills to turn them into reports, executive summaries, infographics, and data stories.

Request Early Access

Request early access at dreambase.com/mcp.

Agents should not have to start over every time

Most agent workflows against company data still begin in the same place: a vague question and a blank map.

An agent has to discover the schema. It has to infer which tables belong together. It has to write a query, inspect the result, correct itself, and try again. The next person asks a similar question tomorrow, and the entire loop starts over.

That is data churn.

It is expensive in every sense of the word. It burns tokens and usage limits. It creates avoidable load on the systems running your product. It leaves room for different people and different agents to arrive at different definitions of the same KPI.

Raw database access does not solve that problem. It gives an agent more to rediscover.

The Dreambase MCP gives it a prepared analytical environment instead.

One MCP, all your data sources

Connect Dreambase MCP once, and it becomes the managed analytical layer between your agents and the data that runs your business.

Dreambase unifies your connected Postgres databases, APIs, and MCP connectors into one governed environment. Rather than asking an agent to reach into each source, learn its shape, and reason through the joins at query time, Dreambase prepares the work ahead of time.

Your agents get:

• Pre-calculated historical metrics and KPIs across daily, weekly, monthly, and benchmark views.

• Validated analytical datasets stored as Parquet outside your production database, with DuckDB available for fast, enterprise-scale analysis.

• Mapped schemas, definitions, and relationships so the agent does not need to guess where an answer lives or how a metric is calculated.

• Managed, OAuth-scoped access to every connected source through one MCP connection, designed for a whole team rather than one person with a direct database credential.

• Skills that know the work: how to build the right insight, visualization, executive report, infographic, or data story for the question at hand.

That changes the question an agent can answer.

Instead of asking, “What tables are here?” it can ask, “What changed in expansion revenue last quarter, which cohort drove it, and what should the executive team do about it?”

Instead of working through a sequence of expensive ad hoc queries, it can use a prepared dataset and a KPI that has already been defined, calculated, and validated.

This is not another pipeline. It is a data platter for every agent you use.

Built for the agents already in your workflow

The Dreambase MCP is designed to meet your team where it works today.

Use it with Claude Cowork and Claude Code to turn business data into high-quality reports, executive summaries, and visual artifacts. Use it in ChatGPT, Codex, or Cursor when your work lives closer to a codebase, a product decision, or an internal tool. Add it to an agentic workflow when an automated process needs reliable metrics without a person opening a dashboard at all.

The output does not have to look like a chart.

It can be a concise weekly decision brief. An infographic for an all-hands meeting. An executive summary that identifies what moved and why. A product analysis attached to a planning document. A prepared answer inside the chat where a team is already having the conversation.

The interface becomes optional. The analytical foundation does not.

Ways Dreambase MCP can plug in your workflow

Beautiful data stories and infographics based on your datasets (shown in Claude Cowork)

Fully custom dashboard using your datasets, KPIs, and custom charts skills from Dreambase (shown in Codex)

Integrate Dreambase MCP tools into your application code for analysis backed by your data (shown in Cursor)

Customized agentic workflows blending Dreambase agents and tools with your own (shown in Claude Code)

Custom Supabase health reports (shown in ChatGPT)

Data & KPI skills are your superpower

Agents can now list and search available Data and KPI Skills, select the right ones automatically, and create new Skills when the work calls for a reusable capability. That matters because the most valuable business knowledge is not only in a database.

It is in the definition of a KPI. The trusted way to join sources. The standard for an executive report. The logic behind a recurring analysis. The visual grammar that makes a data story clear instead of noisy.

Skills turn that knowledge into something an agent can use repeatedly.

Each reusable skill removes another round of prompt-writing, guesswork, token burn, and correction. It gives every team member a more consistent starting point. And it gives Dreambase a way to improve as it learns how your organization actually works, rather than resetting every time someone starts a new chat.

This is how agent capabilities compound.

A more secure and efficient path than direct database MCP

Supabase MCP is an important step forward. It gives agents a direct path into the systems developers already use.

But direct database MCP access is still direct access. The agent must discover and interpret the data at runtime, query against an operational system, and repeat that work whenever the context disappears or a new workflow begins.

Dreambase is designed to reduce that exposure.

The data your agents need is prepared outside production as validated datasets. The most important metrics are computed in advance. Access is managed across the entire team through a single connection. Results are shaped for agent use, so agents can answer the question instead of consuming their context window figuring out how to ask it.

That means fewer unnecessary production queries, fewer raw credentials to distribute, and a more controlled path from source data to business answer.

Faster is good. More consistent is better. Safer is non-negotiable.

The journey is just getting started

We wrote recently that dashboards are not the destination. Headless Analytics is what happens when the trusted data work behind them becomes available to every agent and workflow that needs it.

The Dreambase MCP is the beginning of that vision becoming practical.

It gives your agents the context they need before the question arrives. It gives your team a shared foundation that improves over time. And it gives Dreambase a new surface for turning validated data into whatever form the work demands.

Reports. Executive summaries. Infographics. Dashboards. Agents. Workflows. Interfaces nobody has imagined yet.

This is the next step.

Request early access to Dreambase MCP at dreambase.com/mcp.

Stay tuned.