Learning from your corrections
Confirmations, rejections and manual edits are recorded as learning patterns for the client and practice, so suggestions follow your style.
The problem it solves
AI that does not learn from corrections never gets better.
How it works
- Review AI suggestions.
- Confirm, reject or edit them.
- Future suggestions reflect what you did.
What your practice gets
- Suggestions improve with use.
- Your practice's style.
- Per-client patterns.
Questions accountants ask
Does it learn from manual edits?
Yes.
Is learning per client?
Both per client and across your practice.
Can I undo what it learned?
Rejecting a suggestion teaches it not to repeat it.
Related features
Features that work alongside learning from your corrections.
AI account suggestions
Ask AI to suggest accounts for selected or filtered transactions in a batch, then confirm the suggestions you agree with.
Client memory for the AI assistant
The AI assistant keeps client-specific notes between sessions, with older memories fading over time.
Anomaly detection
Unusual transactions and period anomalies are flagged for you to investigate.
Background AI jobs
Large AI coding, comment and GST jobs run in the background with progress you can check, so you can close the page.
Pre-finalisation review checks
A review pass across categories, comments, GST, duplicates, period anomalies and missing accounts before you finalise.
AI suggestion statistics
See how often AI suggestions are accepted for each client.
See it working in your practice.
Start with one workflow and adopt more of the connected practice system over time. Your team reviews and approves the work.
14-day free trial. A payment method is required to start. Your paid subscription starts automatically when the 14 days end; cancel during the trial and you pay nothing. After the trial, a 6-month minimum term applies, then month-to-month. Commit to 12 months up front and the monthly price is 20% lower. See pricing.