What is an active table?

An active table is a spreadsheet-database hybrid where columns execute work — calling APIs, running AI prompts, computing formulas, and pushing to integrations — automatically for every row.

Traditional tables hold data. Active tables do things. Hypertab coined the term in 2026 to describe a category between databases (passive storage) and workflow tools (event-driven execution). The active table is a third pattern: persistent data with columns that execute logic per row.

Active table vs. database vs. workflow tool

Three distinct categories. Understand the trade-offs.

Pattern	Examples	What columns do	Best for
Database	Postgres, MySQL, Supabase	Hold data only	Transactional apps, analytics
Spreadsheet	Google Sheets, Excel, Airtable	Hold data + compute local formulas	Light data work, ad-hoc analysis
Workflow tool	n8n, Zapier, Make	N/A — execute graph per event	Event-driven automations
Active table	Hypertab, Clay (subset)	Hold data + execute work per row	Batch enrichment, AI ops, GTM pipelines

The eight column kinds in Hypertab

Every column has a kind field that determines its behavior.

static

Holds data

e.g. Name, email, status

Runs an AI prompt per row

e.g. Generate outreach copy from company info

http

Calls an API per row, extracts response

e.g. Hit Clearbit /enrich endpoint, save to row

formula

Computes from other columns

e.g. `{{first_name}} {{last_name}}` → full name

integration

Pushes to external service per row

e.g. Sync to HubSpot when score > 80

waterfall

Tries sources in order, uses first match

e.g. Try Apollo, then Hunter, then ZoomInfo

lookup

Cross-table VLOOKUP

e.g. Match deal to account by domain

extract

Pull a JSON field from upstream column (zero API cost)

e.g. Extract `.company.name` from raw HTTP response

How execution works: row-waterfall DAG

Smart columns don't run in isolation. Hypertab builds a dependency graph from column references ({{col_name}} in prompts, URLs, formulas). Rows flow through this graph in topological order: A → B → C → D, per row, in parallel across rows.

1. Insert row via webhook, MCP, CSV, or REST.
2. DAG fires if any smart column has auto_run: true.
3. Topological layers execute — independent columns run in parallel within a layer.
4. Each cell stores its output — queryable like any database field.
5. Edits cascade — change an upstream cell, dependents recompute.

Plan-gated cross-row parallelism: free 5 → enterprise 200 concurrent row DAGs. Token-bucket rate limits per (provider, account) keep external APIs happy.

When NOT to use an active table

Honest caveats.

Real-time transactional apps: use Postgres or Supabase.
Long-running step functions with branching: use Temporal or Inngest.
Heavy collaboration / docs: use Notion or Coda.
BI dashboards on warehouse data: use Metabase, Hex, or Mode.

FAQ

What is an active table?: An active table is a spreadsheet-database hybrid where columns execute work — calling APIs, running AI prompts, computing formulas, and pushing to integrations — automatically for every row. Hypertab is an AI-native active table, operable by AI agents through the Model Context Protocol (MCP) and by humans through a browser.
How is an active table different from a regular database?: A regular database (Postgres, MySQL, Supabase) stores rows and runs queries. An active table also stores rows, but each column can also DO work: call an API, run AI, compute a formula. Logic lives with the data, not in a separate workflow tool.
How is an active table different from a spreadsheet?: Spreadsheets (Google Sheets, Excel) compute formulas locally. Active tables also call external APIs, run AI prompts, and push to integrations — at table scale, with rate limiting, retries, and dependency graphs. Formulas are one of eight column kinds, not the only one.
How is an active table different from a workflow tool like n8n or Zapier?: Workflow tools execute one graph per event. 50K rows means 50K runs. An active table runs columns at table scale — 50K rows is one batch job. Columns also persist their output as queryable cells, so you can re-run, edit, or audit them like database fields.
Where did the term "active table" come from?: Hypertab coined the term in 2026 to describe a category between databases (passive storage) and workflow tools (event-driven execution). The active table is a third pattern: persistent data with columns that execute logic per row.

Try the active table

Free tier, no credit card. Sign up, paste your MCP config, and your agent operates the table.

get started free →