Recently, I used Microsoft Copilot to complete a documentation-heavy assignment and successfully meet my goal.

The task involved multiple Excel files containing different pieces of information. I needed to connect the data, organize it into a single structure, analyze the results, create visual summaries, and turn the findings into clear documentation.

Normally, this type of work can require a significant amount of manual effort.

Files must be opened individually.
Columns must be compared.
Information must be copied or matched.
Errors and missing values must be reviewed.
The final results must then be converted into a report that other people can understand.

By using Copilot together with Microsoft Excel and Word, I was able to complete the work more efficiently.

However, the most valuable result was not simply completing the task faster.

I also transformed the completed process into a reusable prompt-based guideline, allowing another person to perform the same work in the future.

This experience changed the way I think about AI productivity.

AI is not only a tool for producing an answer.

It can also help us capture the way we work and turn individual experience into a reusable organizational capability.

---

The Challenge: Multiple Excel Files, One Final Result

The starting point was a common documentation problem.

The information I needed was distributed across several Excel files.

Each file contained part of the overall picture. Some data had to be matched using common identifiers, while other information needed to be categorized, summarized, or reviewed for inconsistencies.

Working manually would have required me to:

• open and review each workbook
• compare different column structures
• identify matching records
• consolidate the required fields
• check duplicate or missing data
• calculate the overall status
• create charts or summaries
• prepare the final written documentation

This type of work is especially common in Quality Assurance.

In the medical device industry, QA professionals frequently manage information from multiple sources, including:

• complaint records
• CAPA trackers
• audit findings
• training records
• change controls
• post-market surveillance data
• regulatory submission trackers
• document revision histories

The challenge is often not the lack of data.

The real challenge is turning fragmented data into verified and understandable information.

---

Why Microsoft Copilot Was Especially Useful

One of the strongest advantages of Microsoft Copilot is its close relationship with Microsoft 365 applications.

Excel and Word are already central tools for documentation work. By using Copilot within this familiar environment, I did not need to completely redesign my working process or transfer confidential business information into an unfamiliar system.

Copilot helped me think through the structure of the work while Excel remained the primary environment for data handling.

This combination was particularly useful for three reasons.

First, I could work with the Excel files I already had.

Second, I could use natural-language prompts to support formulas, summaries, comparisons, and visualization.

Third, I could use Word to turn the final analysis into clearer written documentation.

The workflow became:

Multiple Excel files → Consolidated data → Analysis → Visualization → Written documentation

That alone made the work significantly more manageable.

---

Step 1: Connecting Multiple Excel Files

The first stage was to connect several data files into a single working structure.

Instead of treating every workbook as an isolated file, I identified the common fields and created a central dataset.

The purpose was not simply to copy all rows into one large spreadsheet.

The goal was to create a structure that could be:

• filtered
• compared
• updated
• summarized
• verified
• visualized

For example, common identifiers could be used to connect records across files, while status fields, dates, owners, categories, or risk levels could be standardized.

Copilot supported this process by helping me clarify:

• which columns should be compared
• which fields appeared to contain the same information
• how data could be grouped
• what inconsistencies should be reviewed
• which calculations would be useful

This reduced repetitive handling and made the overall dataset easier to manage.

It also reminded me of an important principle:

AI becomes far more useful when the underlying data is clearly structured.

Copilot could assist with the work, but the quality of the output still depended on the quality of the source data and the logic used to connect it.

---

Step 2: Visualizing the Consolidated Information

After combining the data, the next step was visualization.

This was one of the most valuable parts of the process.

When information is spread across many Excel rows, it can be difficult to understand the overall situation. Once the data is summarized visually, patterns become easier to recognize.

The visualization helped answer practical questions such as:

• How many items were complete?
• How many remained open?
• Which tasks were overdue?
• Which owner had the largest number of open actions?
• Which category contained the most issues?
• Were there repeated patterns across the data?
• Where should follow-up efforts be prioritized?

I used Copilot to help organize the information into useful summaries and visual outputs.

These could include:

• completion-rate charts
• open-item summaries
• overdue-item trends
• category comparisons
• status dashboards
• owner-based workload summaries

The visualization was not merely decorative.

It turned the spreadsheet into a decision-making tool.

This distinction is important:

Raw data tells us what exists.
Visualization helps us understand what requires attention.

---

Step 3: Completing the Written Documentation

Once the Excel data had been organized and reviewed, I used the results to complete the written documentation.

The consolidated dataset provided the factual foundation.

The visualization provided the overall picture.

Word then became the environment for presenting the purpose, methodology, results, key findings, and next actions in a structured way.

Copilot helped me convert analytical outputs into more readable documentation.

For example, it could support the preparation of:

• executive summaries
• methodology descriptions
• result summaries
• issue descriptions
• action-item lists
• management updates
• process guidelines

However, I did not simply accept the generated language.

I reviewed the wording, compared it with the source data, and corrected anything that was unclear or unsupported.

This is particularly important in QA and medical-device work.

AI can support documentation, but the responsible professional must still verify:

• factual accuracy
• source traceability
• dates and quantities
• regulatory terminology
• risk statements
• conclusions
• approval requirements

Copilot helped me complete the task more efficiently.

It did not replace my responsibility for the final result.

---

Step 4: Asking Copilot to Work Backward

After completing the assignment, I asked Copilot to do something different.

Instead of generating another output, I asked it to help me reconstruct the process I had followed.

I wanted Copilot to analyze the completed workflow and convert it into a set of reusable prompts.

The logic was simple:

1. Review the finished work.
2. Identify the major steps.
3. Determine what information was required at each step.
4. Reconstruct the instructions that produced the result.
5. Rewrite those instructions as reusable prompts.
6. Organize the prompts in the correct sequence.

This was the turning point.

Until that stage, Copilot had helped me perform one assignment.

After that stage, it helped me document how the assignment could be repeated.

The workflow was no longer held only in my memory.

It became a reusable method.

---

Step 5: Turning the Prompts into a Guideline

I then placed the reconstructed prompts into a practical guideline.

The guideline was designed so that another person could understand:

• the purpose of the work
• the required input files
• the fields that should be reviewed
• the sequence of activities
• the prompts that should be used
• the expected output from each prompt
• the manual verification points
• the final documentation requirements

This created a form of AI-assisted work instruction.

Traditional procedures often explain what employees must do.

A prompt-based guideline can go one step further by also providing a structured way to interact with AI during the work.

For example, instead of writing only:

Consolidate the source files and review missing data.

The guideline could include a prompt such as:

Review the available Excel tables and identify columns that represent the same type of information. Propose a consolidated table structure using the record ID as the primary matching field. List unmatched records, duplicate identifiers, and inconsistent data separately for manual review.

This gives the next user much more practical support.

It explains not only the required outcome, but also how Copilot can help reach that outcome.