> ## Documentation Index
> Fetch the complete documentation index at: https://docs.asteragents.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Sage Intacct

> Connect Sage Intacct to your Agents for read-only access to your accounting data

## Overview

The [Sage Intacct](https://www.sage.com/en-us/sage-business-cloud/intacct/) integration gives your agents **read-only** access to your Intacct company. Agents can query GL detail, AP bills, AR invoices, vendors, customers, the chart of accounts, and any other Intacct object visible to the connected user — to answer financial questions, reconcile records, and feed reporting, without ever modifying anything in Intacct.

The integration is read-only by design for the initial release: the tools only issue Intacct read functions, so agents cannot create, update, or delete records.

## Prerequisites

* A Sage Intacct company with **Web Services** enabled
* A **Sender ID** and **Sender Password** (the application-level Web Services credential issued by Sage — your Sage account team or implementation partner provides this)
* A **Web Services user** in your company (Company > Web Services users) — a dedicated user with read-only permissions on the relevant modules is recommended
* An Aster **admin** to connect the integration (the connection is shared org-wide)

## Setup Guide

<Steps>
  <Step title="Gather credentials">
    You need five values: the **Sender ID** + **Sender Password** (from Sage), and your **Company ID**, **Web Services User ID**, and **User Password**. Multi-entity companies can optionally pin the connection to a single **Entity ID**.
  </Step>

  <Step title="Connect Sage Intacct in Aster Agents">
    Navigate to **Control Hub > Integrations**, locate the **Sage Intacct** card, click **Connect**, and enter the credentials. Aster validates them live against the Intacct gateway and confirms which company you connected. Sessions are managed automatically from there.
  </Step>

  <Step title="Add Tools to Your Agent">
    Edit your agent and enable **Sage Intacct Get Schema** and **Sage Intacct Query**. The agent discovers the available objects and fields at runtime, then queries them.
  </Step>
</Steps>

## Available Tools

* **Sage Intacct Get Schema** (`intacct_get_schema`) — Lists every Intacct object visible to the connected user, and returns exact field names, types, and relationships for any object. Agents call this first instead of guessing field names.
* **Sage Intacct Query** (`intacct_query`) — Reads records from any Intacct object with a SQL-ish filter predicate (e.g. `TOTALDUE > 1000 AND STATE = 'posted'`). Auto-paginates up to a configurable row cap; **large result sets spill to a CSV attachment** while the inline payload is truncated to a token budget, so big pulls don't balloon the agent's context.

<Note>
  **Read-only.** This release exposes reads only — agents cannot create, update, or delete Intacct records. Write capability is intentionally not wired up.
</Note>

## Example Prompts

Once connected, agents can answer questions like:

* "List all open AP bills over \$10,000 with their due dates."
* "Pull GL detail for account 6000 for May 2026." (the full set spills to a CSV the agent can process in Python)
* "Which vendors did we onboard this quarter?"
* "Reconcile AR invoices against this customer remittance file."

## Notes & Limitations

* **Read-only** — no writes in this release, by design.
* **Permissions follow the connected user** — agents only see the objects and entities the Web Services user can see. Scope that user's roles accordingly.
* **Filter syntax** — `intacct_query` uses Intacct's query predicate syntax (`=`, `<`, `>`, `like`, `in`, `and`, `or`; dates as `MM/DD/YYYY`), not full SQL — there is no ORDER BY or JOIN.
* **Page size** — Intacct pages at 1,000 rows server-side; the tool auto-paginates to reach larger totals.
* **Usage metering** — Intacct meters Web Services calls per month by performance tier. Each page of a large pull is one call; prefer narrow filters over bulk pulls on transactional objects.

## Security Considerations

* **Org-level connection** — one admin connects once for the whole organization. Passwords and session tokens are stored server-side and redacted when the connection is read back in the UI.
* **Read-only by construction** — the tools cannot mutate your Intacct company.
* **Use a dedicated Web Services user** with least-privilege, read-only roles, so access can be audited and revoked independently.
* **Grant the Intacct tools only to agents that need them.**

## Troubleshooting

<AccordionGroup>
  <Accordion title="Connection rejected on setup">
    * The error message says which credential pair failed: **sender authentication** means the Sender ID/Password; **login failed** means the Company ID / User ID / Password.
    * Confirm the user is a **Web Services user** (not just a regular login) and is not locked out.
    * Confirm your Sender ID is authorized for your company under **Company > Company Info > Security > Web Services authorizations**.
  </Accordion>

  <Accordion title="A query returns no rows or an unknown-object error">
    * Run **Sage Intacct Get Schema** to confirm the exact object and field names — they're case-sensitive uppercase identifiers (e.g. `GLDETAIL`, not `GLDetail`).
    * Loosen the `query` predicate; remember dates are `MM/DD/YYYY`.
    * Check the connected user's module permissions — objects they can't read don't appear at all.
  </Accordion>

  <Accordion title="Large pull feels slow or truncated inline">
    * Intacct pages at 1,000 rows, so big pulls take several sequential calls. The full set is written to a CSV attachment; the inline preview is intentionally truncated. Have the agent process the CSV (e.g. in `execute_python`) rather than reading every row inline.
  </Accordion>
</AccordionGroup>

## Related Tools

* [Run Code](/tools/execute_python) — process large Intacct result sets (the CSV spillover attachment) with pandas.