Inform Users of Data Filled By AI in Invoice Capture

[DawnM]

Has anyone run into the problem of errors populating fields and confusing users? Even the wrong vendor has slipped passed multiple reviews on a couple occasions.

Is there any way to turn off the AI function, or at least bring attention to the selection made by AI with highlighting of some sort?

It is like having an employee that is untrainable. At least for our small business, we do not have enough volume of similar invoices to benefit from it.

[KevinD]

@DawnM are you using our Invoice Capture service? 

Could you please provide some examples of the errors? 

[DawnM]

@KevinD Yes, we are using Invoice Capture service. It is in that stage that the fields are populated, but not by automation.

The worst is the vendor error. A vendor that this happened with recently was a city park, but there is a couple vendors that it consistently will not recognize due to the placing of the vendor name in the Invoice heading. It will still make a selection, just not the right one.

The GL/Expense code is pretty annoying too. It will often choose the most recent code selected from the previous invoice, which rarely is correct.

[gptdeutsch]

I understand your concern regarding AI populating fields and potentially causing errors in your invoice capture process. While turning off the AI function entirely might not be an option, there are some steps you can take to address this issue.

1. Review and Correct: One approach could be to thoroughly review the invoices generated by AI. After AI completes its task, have a human reviewer go through the data and cross-check it with the original documents. This additional step can help catch any errors or discrepancies.

2. Customization: Check if your AI solution allows for customization. You may be able to train the AI to better understand your specific invoice requirements over time, reducing errors.

3. User Training: Consider providing training to your employees who use the system. Educate them on how the AI works, its limitations, and how to identify and correct errors.

4. Feedback Loop: Establish a feedback loop where users can report any issues or errors encountered in the process. This can help in improving the system's accuracy over time.

5. Highlighting: Request a feature enhancement that would allow the AI-generated data to be highlighted or tagged for easy identification by users. This could serve as a visual cue for manual review.

6. Volume Increase: As your business grows, you may reach a point where the AI's accuracy improves due to a higher volume of invoices. In the meantime, complement AI with human oversight.

Remember that AI systems tend to improve with time and data. While it might not be perfect initially, with the right adjustments and feedback, it can become a valuable tool in your workflow.