Checklist: AI Office & Hanna Excel Reporting

19/10/2026

Half the Time, and the Trap Behind That Number

When speaking with department heads in Vietnam, I often hear them estimate: "About half of our weekly working hours are spent aggregating Excel data." It sounds reasonable, but this figure is easily misleading. It leads managers to believe that buying an AI tool will automatically recover 50% of that productivity. In reality, AI does not create data. It only processes what you already have at speed. If the input data is garbage, Hanna (the AI assistant in AIV Office) will return garbage reports at a higher speed. The issue isn't whether the software is fast enough, but whether you have prepared the data sufficiently for the AI to understand. This is where AIVISION, with its practical deployment experience, typically sits down with clients to review workflows before enabling automation.

Four Critical Decisions That Determine Report Quality

Before using the office AI assistant to consolidate reports, you must make the right choices at the following four points. Each involves a trade-off between short-term time and long-term accuracy.

• Standardize column formats. This is the most important step. The Sales department's Excel uses "Revenue," Accounting uses "Revenue," and Retail uses "Sales." Hanna needs a clear mapping. If you don't standardize, the AI will have to guess, which consumes processing time and carries a high risk of error. Spend the first hour unifying a single column name across the entire system.
• Identify "trusted" data sources. You have data from three different suppliers for the same metric. Which source do you choose as the baseline? Without a priority rule (e.g., using ERP system numbers as the standard and employee-entered Excel numbers as reference), the final report will be contradictory. Clearly document this rule for Hanna to reference when encountering duplicate data.
• Define "clean data." Is a blank cell an error or a zero? Is February 31st an error or a data entry mistake? The AI needs to know how to handle exceptions. If you don't define this, Hanna might skip that data row, resulting in incomplete reports that go unnoticed. List 5 common error cases and decide how to handle each.
• Manage file access permissions. Consolidated reports often contain sensitive data from multiple departments. Who has permission to view the original Excel file? Who only sees the final report? This relates to security within AIV Office. If you don't manage permissions from the start, you will face a data crisis after the report has spread through internal chat groups.

Frequently Asked Questions

Can Hanna automatically fix typos in product names?

It can, but you shouldn't rely on it completely. AI can identify common spelling variations, but for industry-specific product names, one mistake means one incorrect report. Use this feature for alerts, not for automatic replacement, unless you have manually verified it.

Do I need to know complex Excel formulas to use it?

Not necessarily. You only need to describe the desired result in natural language. For example: "Sum revenue by month and divide by the number of orders to get the average order value." Hanna will handle the logic. However, if the formula is too complex, visually checking the results is still mandatory.

How long does it take to set up this workflow?

With standardized data, it takes about 2-3 working days. With fragmented data from multiple sources, it can take up to 1-2 weeks. This time is primarily spent on data cleaning and rule setting, not programming.

Applying the Checklist in Monday's Meeting

Returning to the "half the time" figure at the beginning. If you start the new week with standardized data and clearly defined rules, the time spent on "data aggregation" will decrease significantly, potentially to just one-third of what it was before. Instead of waiting for files from colleagues and checking every cell, you open AIV Office, command Hanna to consolidate, and use the saved time to analyze trends. That is the true value of an office AI assistant: not doing everything for you, but clearing away the trivial tasks so you can focus on decision-making. Bring this checklist to your next Monday meeting and test it with a small report before applying it to the entire department.

Every company hits this differently, and the hard part is usually the data rather than the model. To pressure-test your case quickly, talk to AIVISION - or first see how we deploy and what we have written before.