Zum Hauptinhalt springen

A Collaborative Business Data Platform for Humans & AI agents

What ZeitMind Really Is​

ZeitMind is a managed collaborative PostgreSQL UI designed to make business data accessible to both technical and non-technical audiences — and critically, to their AI agent equivalents.

Rather than a black-box data platform, ZeitMind is built as an open standard API meeting point where business users, builders, and state-of-the-art AI agents can work hand-in-hand on the same data, with the right guardrails in place.

The Core Architecture​

Each ZeitMind project gets:

  • A dedicated PostgreSQL database with unique credentials
  • No lock-in — access all your data with open-source postgres tools
  • Collaborative workspace with sharing/permissions, schedules, and data pipelines

This opinionated setup gives you all the essentials needed to build internal tooling and connect to business data, without the complexity of stitching together separate systems.

The AI Collaboration Model​

ZeitMind brings together two types of users — and two types of AI agents:

Human Users​

  • Operational end users — business analysts, managers, stakeholders
  • Semi-technical builders — data analysts, internal tool creators

AI Agents​

  • Builder Agent — creates tables, writes SQL queries, designs charts, proposes edits to data pipelines
  • Operator Agent — operates applications like a user (inspecting panels, setting inputs, triggering actions) to automate tedious tasks like extensive drill-downs or product trend exploration

The Guardrails Problem​

Working hand-in-hand with AI agents on live business data is extremely powerful — but also dangerous:

  • Agents could mess up business logic
  • Agents could delete critical data
  • Agents could leak sensitive information

This is where ZeitMind's engineering focus lies: putting up the right guardrails to make AI-human collaboration non-harmful, effective, and constructive.

Safety Measures​

  • Query parsing and validation
  • Permission systems and access control
  • Data branching (coming soon) — test changes without affecting production
  • Audit trails and rollback capabilities

The Data Pipeline Workflow​

ZeitMind's reactive data pipeline system ensures repeatability and automation:

  1. Raw data → ingested from external sources or uploaded files
  2. Derived tables → SQL transformations, cleaning, parsing, combining
  3. Downstream dependencies → when you save a query, all downstream tables re-materialize automatically
  4. Applications & analyses → built on top of reliable, versioned data

This graph-based approach means when fresh data arrives weekly or monthly, your entire pipeline runs automatically — from raw to report.

The Collaborative Builder Agent​

The Builder Agent lives alongside you in the workspace:

  • Spawn via sidebar (+ next to Analyses)
  • Creates Artefact Tables — temporary tables that can be used in apps or promoted to permanent derived tables
  • Proposes edits to derived table queries for your review/acceptance
  • Helps build complex pipelines — parsing, cleaning, imputing, combining data step-by-step

When you accept an agent's edit, the system materializes the table and cascades changes downstream — ensuring consistency across your entire data model.

The Operator Agent (Advanced)​

The Operator Agent works at the application layer:

  • Inspects panels and UI components
  • Sets inputs and parameters
  • Triggers actions to navigate and drill down

Use it to automate tedious tasks like:

  • Extensive drill-downs into sub-accounts when forecasts aren't met
  • Exploring specific products and their trends to surface decisions requiring human judgment

Why Open Standards Matter​

ZeitMind doesn't lock you into proprietary formats or APIs:

  • PostgreSQL backbone — industry-standard, battle-tested, universally supported
  • External table registration — connect to your existing databases (credentials accessible via 3-dot menu in Data Prep)
  • Transparent query execution — see exactly what SQL is running, preview before saving

This means you can:

  • Connect your own tools and scripts
  • Export data anywhere
  • Migrate away if needed (though we hope you won't want to!)

The Vision​

ZeitMind is purpose-built for the era where AI agents are teammates, not just tools.

By creating a structured, safe environment where humans and agents collaborate on the same data — with the right permissions, validations, and rollback mechanisms — we're enabling a new class of internal tooling that's more powerful, more accessible, and more trustworthy than what came before.

In short: ZeitMind is managed PostgreSQL for the AI age — opinionated, collaborative, and built with guardrails that make human-AI data work safe and effective.