AppsAutessaDB

Semantic search and AI · AutessaDB

Database tools for AI agents

AutessaDB is a backend for AI agents. Ready-made AutessaDB tools let an agent insert, update, delete, bulk write, query, aggregate and run semantic search on a table, browse and grep a Drive, and query the data lake with SQL, with no connector to set up. Agent writes go through the same pipeline as people's edits, so triggers fire and computed columns recompute.

Goes beyond a typical managed database
SupportAgent · conversationruns as: dana@SAMPLE
How many open tickets mention refunds?
tool · AUTESSADB_SEMANTIC_QUERYtickets.body · “refund” · limit 10 · 7 matches
tool · AUTESSADB_AGGREGATECOUNT where status = open · 5
Five open tickets ask about refunds. The oldest is #2201, waiting 3 days.

-- the problem

Agents need memory and records, safely

An agent that cannot write anything forgets; an agent with raw database credentials is a risk. AutessaDB gives agents precise tools on precise tables, running as a known identity, so every read is filtered by row rules and every write is validated and logged.

What you get

  • Write tools: AUTESSADB_INSERT, AUTESSADB_UPDATE, AUTESSADB_DELETE and the bulk variants.
  • Read tools: AUTESSADB_QUERY_TABLE (page size 50, capped at 100, with cursors) and AUTESSADB_AGGREGATE.
  • Search tools: AUTESSADB_SEMANTIC_QUERY returns scored JSON; AUTESSADB_GENERAL_SEARCH returns readable markdown tuned for precision (topK 5, threshold 0.7).
  • AUTESSADB_DRIVE gives ls, tree, find, stat, grep, read, search and text writes on a Drive.
  • Analytics tools describe tables, run SQL and fetch results; Create agent tools makes all three in one click.
  • Tool configs can fill values from agent inputs with @@{VARIABLE} placeholders.
  • Query results include the schema, so the agent can read column types without another call.

-- how it works

How does database tools for ai agents work in AutessaDB?

  1. Create the tool

    Pick an AutessaDB tool type and bind it to a table, with fixed or @@{} variable values.

  2. Attach it

    Give the tool to an agent alongside its instructions.

  3. Run as someone

    The agent acts as the person or key it runs as, inside their row and column rules.

  4. Watch it work

    Writes fire triggers and land in the access log like any other write.

-- every option

AutessaDB tool types

optionwhat it does
AUTESSADB_INSERT / UPDATE / DELETESingle-row writes with typed values, or deletes by filter.
AUTESSADB_BULK_INSERT / UPDATE / DELETEMany rows per call.
AUTESSADB_QUERY_TABLEFiltered, sorted, paged reads with cursors.
AUTESSADB_AGGREGATEGrouped COUNT, SUM, AVG, MIN, MAX.
AUTESSADB_SEMANTIC_QUERYVector search with scores and paging, as JSON.
AUTESSADB_GENERAL_SEARCHVector search as readable markdown.
AUTESSADB_DRIVEFile operations on a Drive.
Analytics describe / query / get resultSQL over the data lake.

-- example

A semantic search tool for RAGjson
{
  "toolType": "AUTESSADB_SEMANTIC_QUERY",
  "tableId": "V1StGXR8Z5",
  "query": "@@{SEARCH_QUERY}",
  "columnName": "document_text",
  "limit": 10,
  "scoreThreshold": 0.5,
  "page": 1,
  "includeDetails": true
}

Full reference in the AutessaDB CLI docs and the API reference.

-- permissions

Who can do what?

accesscan
Tool builderCreate tools on tables they can access.
Agent at run timeExactly what its acting identity can see and change.

◆ AI assist

Where AI helps

Agents choose when to call each tool. For the data lake, they describe tables first, prefer verified queries, filter on partition columns and page through long results.

AI in AutessaDB works inside the same access rules as the person using it, and schema changes always wait for a person to apply them.

-- relations

How it connects

autessa platform

AI agents

Attach AutessaDB tools to any agent.

autessa platform

Scripts and tools

Combine table tools with your own script tools.

autessadb feature

Row-level security

Agents never see more than their identity.

autessadb feature

Semantic layer

Helps agents write correct SQL.

-- faq

Questions about database tools for ai agents

Do agents need a connector to use AutessaDB?

No. AutessaDB tool types take a table id and work directly.

Can an agent see rows a user cannot?

No. Agents act as the person or key they run as, and every read and write passes that identity's row and column rules.

Do agent writes fire triggers?

Yes. Agent writes go through the same pipeline as people's edits, so triggers fire and computed columns recompute.

$ autessa autessadb tables create

Try database tools for ai agents in AutessaDB

Ready-made tools give agents insert, update, delete, query, aggregate, semantic query, general search, Drive and data lake access to your tables.

Read the AutessaDB guide or install the CLI.

100% customizable, code and all

Make AutessaDB fully your own.

Use it out of the box, then change anything. Ask Autobot, Autessa’s AI builder, for a new field, screen, report, workflow or agent and it builds it into your copy of the app. Or open the code and edit it yourself. Nothing is locked: the app, its data and its code are yours.

Start free and make it yours