Xero AI and MYOB AI for Accountants

Xero AI uses JAX to answer questions and reconcile transactions. MYOB’s Iris features explain business figures and help prepare BAS, with different permissions and review steps.

What Xero AI Does

JAX is Xero’s AI finance partner. Its chat answers financial questions and helps create invoices and quotes. Automatic bank reconciliation works in the background once you enable it for a bank account.

Xero’s JAX documentation says chat is available to all subscribers and users without an additional charge. The information a person can access follows their Xero permissions. Available actions also depend on the organisation’s plan.

For Australian subscriptions, automatic reconciliation is included with Grow, Comprehensive and Ultimate. Ignite can add it through AI Automate. These are separate entitlements from opening JAX chat.

When statement lines arrive, JAX uses bank rules, existing transaction matches, previous coding and predictions to reconcile high-confidence items. Other items remain for review. Xero’s reconciliation guide explains the matching methods and plan access.

The distinction affects your review queue. A suggestion waits for acceptance, while an automatically reconciled item has already changed the books. Review both populations, including the items that disappear from the unreconciled list.

Treat an invoice or quote created through JAX as work to review before sending. Confirm the customer and amounts against the instructions you received. Apply the same check to any client message drafted with AI, particularly its explanation of outstanding payments.

MYOB AI Features

MYOB uses Iris for several different features. Ask Iris answers questions, Smart Reconciliation matches transactions and AI BAS checks BAS preparation. An accountant needs to distinguish a read-only answer from a suggestion that updates the books.

Ask Iris, previously AI Assistant, is available in Business Lite, Pro and AccountRight browser. It explains figures such as income and outstanding invoices. It cannot change the file or settings.

MYOB’s AI Business Insights supplies financial summaries for eligible Pro and AccountRight browser customers. Smart Reconciliation is progressively rolling out. Its matching uses existing records and prior allocations, with predictions based on similar transactions in other MYOB businesses.

The MYOB matching documentation explains how to see the method behind a match. Use that detail to identify a rule that repeatedly applies the wrong treatment.

AI BAS

MYOB AI BAS flags missing documents and GST issues. Accepting a suggestion updates the books. The accountant or business then reviews the result and lodges through their usual method.

AI BAS is available to GST-registered, non-employing and employing Australian businesses using Business Lite or Pro. It supports simple quarterly cash-based BAS (G1, 1A, 1B), plus PAYG withholding (W1/W2) for employing businesses.

AI BAS is an additional Premium Feature with an annual commitment billed monthly.

MYOB Acumatica

MYOB Acumatica, formerly MYOB Advanced, has an ERP-focused AI offering. It includes bill and expense recognition, anomaly detection and AI Automation. These features answer a larger business’s operational needs, beyond a small-business ledger assistant.

Its Service Terms distinguish the read-only AI Assistant from AI Automation, which can run user-initiated actions. Administrators select the Generic Inquiries that expose ERP records to the assistant. Feature access and capacity follow the customer’s entitlements.

For a practice, compare the work in each client file. Xero’s reconciliation entitlement centres on bank transactions. MYOB adds a named BAS-preparation feature with its own eligibility. Acumatica needs an ERP-specific assessment of permissions and actions.

What the AI Sees in Client Books

Ledger AI uses the records available to its feature and the signed-in user. That can include bank descriptions, transaction amounts and contact names. Information you type into chat adds to what the feature receives.

Xero says JAX chat does not use your data to train its language models. Its providers process prompts, which can contain personal information, under arrangements that prevent model-training retention.

Automatic reconciliation has different wording. Xero’s reconciliation data statement permits training where customer data is anonymised or permission has been given. A statement about JAX chat cannot be applied to every Xero AI feature.

Xero’s terms grant data-use rights for operating and improving services. They also allow anonymised statistical data to be used for product development. Keep those contractual permissions separate from the narrower chat assurance.

Ask Iris documentation says business data is not used to train AI models. That assurance concerns Ask Iris. MYOB Business’s terms still make the customer responsible for checking AI-generated information before relying on it.

MYOB Acumatica’s Service Terms expressly permit MYOB to use customer data, prompts and outputs to train or improve its AI features. MYOB’s statement that data does not train external models is a narrower claim about external providers.

Acumatica can process selected ERP records, free-text fields and conversation context. Its terms allow processing outside Australia or New Zealand. Configure approved inquiries to expose only what staff need, and keep unnecessary TFNs and bank details out of prompts.

Before enabling an action for a client, record the feature, its data permissions and the client authority your practice holds. Keep the applicable agreement with that decision. User access remains a separate check from the model’s training policy.

Checks Before Accepting AI Suggestions

Review AI coding against the source document and the client’s circumstances before using the resulting BAS figures. A high-confidence match describes the software’s prediction. It does not establish the business purpose or GST treatment.

1. Confirm the entity, bank account and reporting period. Identify which transactions were automatically processed and which still wait for acceptance.
2. Open the source document for unusual or material items. Compare its supplier and amount with the proposed match. Check for duplicated bills and transfers between the client’s accounts.
3. Review the account and GST code. Investigate private use, mixed purchases and capital items where the bank description alone cannot establish the treatment.
4. Check completeness. Reconcile bank balances, resolve missing documents and investigate uncategorised items. A dashboard percentage cannot establish that every transaction has reached the file.
5. Trace BAS totals to the reviewed transactions using the correct reporting basis. Check any PAYG figures separately where they apply, and retain the client’s approval before lodgment.

For a hypothetical example, a client pays $1,100 for equipment that also has private use. Matching the payment to an equipment account does not settle the business-use proportion. Inspect the invoice and obtain the client’s explanation before accepting the GST treatment.

After correcting an AI suggestion, rerun the affected reports. Save the reviewed BAS summary with the supporting workpapers. If the correction changes an already lodged period, assess the adjustment or amendment before accepting further suggestions for that period.