AutessaDB Tables
AutessaDB is the platform's built-in database. Create tables with typed columns for any structured data — a CRM, an order log, a content pipeline — or knowledge bases optimized for document storage and semantic search. Agents read and write the same tables you edit by hand, and triggers turn row changes into automated pipelines.
Why AutessaDB is different
It looks like a spreadsheet-simple table editor, but three capabilities set AutessaDB apart from an ordinary data store.
AI-searchable
Flip vectorization on for any text, file, or audio column and every insert and update is indexed automatically in the background — no re-index jobs, no separate pipeline. And it's more than raw text going into vectors: an LLM enriches each value with summaries, key facts, and the questions someone would likely ask of it, so search matches meaning rather than exact words. File columns get an auto-generated companion notes column describing each document, and audio columns get a transcript column. Semantic search then works alongside structured queries — the same table answers "which calls mention pricing complaints?" and status = 'open'. Knowledge bases are a one-click preset for bulk file ingestion, but search works on any vectorized column of any table, not only knowledge bases.
Event-driven
Any insert, update, or delete can fire triggers that invoke tools, agents, or multi-agents — or write rows into other tables, whose own triggers run the next step. That's how tables chain into full processing pipelines without queues or glue code. Before you commit to a change, a design-time impact analyzer maps the trigger chain a database event would set off and warns you about potential cycles.
Everything recorded
Every row carries created and last-updated timestamps. Every trigger execution is logged with encrypted before/after payloads you can drill into from the Trigger Execution History. And per-column indexing metadata shows exactly what was vectorized, and when — so you always know what your search index reflects.
Navigating here
From the dashboard sidebar, open AutessaDB → Tables. The list shows all tables with ownership tabs (All Tables / My Tables / Company Shared / Shared With Me), search, and tag filters. The New menu offers four types — Table (define a custom schema), Knowledge base (upload documents for semantic search), Drive (a shared file drive with folders and sharing), and Analytics table (columnar storage queried with SQL) — and the Autobot button opens the DB Copilot. A type badge on each table shows which kind it is, and the list can be filtered by type.
Key concepts
Tables and system columns
A table is a named schema of typed columns. Table and column names use snake_case. Every table automatically gets three read-only system columns managed by the platform: id (integer), created_at, and last_updated.
Column types
The column editor offers these types:
| Type | Holds |
|---|---|
| Short Text | Text up to 32K characters |
| Long Text | Text up to 128K characters |
| File | Documents — PDF, DOCX, XLSX, TXT, and similar |
| Image | PNG / JPG uploads with thumbnail preview |
| Video | MP4, AVI, MOV, and other video formats |
| Audio | MP3, WAV, M4A with an inline player |
| Integer | Whole numbers |
| Decimal | Floating-point numbers |
| Boolean | True/false values |
| Single Select | One value from a defined option list |
| Multi Select | Multiple values from a defined option list |
Per-column options include a Required flag (not available on media columns), min/max length or value constraints for text and numbers, an option-list editor for the select types, and on Integer columns a foreign key link to a row in another table.
Store timestamps as Short Text containing UTC ISO-8601 strings (yyyy-MM-ddTHH:mm:ssZ). The only DateTime columns in a table are the system-managed created_at and last_updated; user-defined date fields follow the Short Text convention so agents, triggers, and scripts can parse them consistently.
Semantic search and vectorization
Mark a Short Text, Long Text, File, or Audio column as Semantic Search to turn on the background indexing described above. The table then gains a working Semantic Search tab: enter a query, pick the vectorized column, and get ranked results with similarity scores. Search agents (see Agents) query these columns at runtime.
Knowledge bases
A knowledge base is the one-click preset for bulk file ingestion: a table pre-configured for document storage and semantic search. The creation wizard lets you drag-and-drop files or folders; they're uploaded, indexed, and immediately searchable. A knowledge base's table view includes a Create Agent shortcut that provisions a ready-made Search agent over the table.
Computed columns
A column's default value can be Static or Computed. Computed values run an AutessaScript snippet: read sibling values with variables.get('column_name') and set the result with response.setResponse(...). Dependencies are detected automatically from the columns your script reads. See Computed Columns for the scripting reference.
DB Copilot (Autobot)
The Autobot button opens an AI copilot that manages your database from natural language — describe what data you need to store and it proposes table creations and schema updates as reviewable change cards you explicitly apply or reject. See DB Copilot.
How to create a table
- Go to AutessaDB → Tables, click Create, and choose Create Table.
- Table Info — name the table in
snake_caseand describe it. - Columns — add each column with a name, type, and description. Set Required, constraints, select options, semantic search, or a default value as needed.
- Review — confirm the schema summary and click Create Table.
To change the schema later, open the table's Manage → Update Schema dialog, which supports adding columns, deleting columns, and appending options to existing select columns.
How to create a knowledge base
- Click Create → Create Knowledge Base.
- Name it, then drag in the documents (files or whole folders) on the Upload Files step.
- Review and create — files are indexed automatically and you land on the table view, ready for the Semantic Search tab or the Create Agent shortcut.
How to edit rows
- Open a table and use the Browse Data tab. Click Add Row to insert — each column gets a typed input, media columns get an upload dropzone with preview, and columns with defaults offer Use Default / Use Computed Value toggles.
- To edit or delete an existing row, right-click it in the grid and choose View/Edit Record or Delete Record.
- Use the column-settings gear to hide, show, and reorder grid columns; media cells offer per-file Download buttons.
Sharing and access levels
Every table starts private to its owner. Share it with specific teammates — each at their own access level — or company-wide at a single level for everyone. The levels carry table-specific meanings:
| Level | What it lets a teammate do |
|---|---|
| READ | View the schema and read rows — get, query, join, and semantic search |
| WRITE | Insert, update, and delete rows |
| PUBLISH | Add triggers to the table |
| ADMIN | Change the schema, share the table, and delete it |
PUBLISH does not grant row writes on a table — writing rows takes WRITE or ADMIN. And while PUBLISH can add a trigger, editing, deleting, enabling, or disabling one requires ADMIN — see Triggers.
Deletion is two-step: the delete request returns a short-lived confirmation token, and confirming re-checks ADMIN. The owner always has ADMIN implicitly; transfer of ownership is owner-only. Tables auto-created by a connector can't be shared, transferred, or deleted directly — manage the connector instead.