Microsoft Copilot: How to Make It Work

Welcome to This Week’s Issue!

This is the second and final deep dive into Microsoft Copilot. After exploring its capabilities, talking to executives, and testing it in real-world scenarios, I can confidently say that Copilot is only truly valuable if you’re fully integrated into the Microsoft ecosystem.

Compared to ChatGPT and Claude, Copilot underperforms in consistency, reasoning, and creativity. However, because so many companies run their operations on Microsoft 365, Copilot will inevitably become the default AI tool for many teams—whether it’s the best option or not.

With the right strategy—structured implementation, adequate training, and some clever tips—teams can enhance Copilot’s accuracy and reliability. In this issue, I will outline some common issues users face, ways to overcome them, and the steps you can take to optimize Copilot’s effectiveness in your daily tasks.

🚀 Big News: We’re Building AI Copilots for Finance!

After months of searching for the right tech partners, I’m excited to announce that we’re developing a suite of AI Copilots for finance professionals. These copilots are designed to eliminate repetitive tasks, optimize workflows, and improve financial decision-making.

To make sure we focus on the right solutions, I’m asking for your input:

✅ Tell me which finance tasks eat up the most time in your day.
✅ Share whether you’ve used AI in your workflow already.
✅ Get the chance to have an AI agent built for you—for free!

➡️ Take 2 minutes to answer the poll

If you can, please share it with your peers—the more insights, the better! Thanks for being part of this journey. 🚀

The Balanced View: Microsoft Copilot – Gaps, Shortcomings, and Mitigation Strategies

Microsoft Copilot, despite its deep integration with the Microsoft Suite, has drawn significant criticism from users across different industries. While it offers seamless connectivity with tools like Excel, Outlook, Word, Teams, PowerPoint, and SharePoint, its performance in real-world applications often falls short of expectations. Below, we outline the most pressing challenges users have encountered with Copilot and strategies to mitigate these issues.

Known Gaps and Shortcomings in Microsoft Copilot

1. Functionality and Reliability Issues

• Reduced Capabilities Over Time: Many users report that Copilot now refuses tasks it previously handled, such as rewriting text or answering basic factual questions.

• Frequent Errors in Core Features: Meeting summaries often omit key action items, Excel formula suggestions reference incorrect cells, and document generation relies too heavily on generic templates, requiring extensive manual edits.

• Lack of Context Retention: Copilot does not maintain continuity between chat sessions, requiring users to re-explain project parameters repeatedly.

2. Usability and Interface Frustrations

• No Option for Message Editing or Response Regeneration: Unlike other AI tools, Copilot does not allow users to refine responses, forcing them to restart interactions.

• Inconsistent Browser Integration: Users frequently encounter broken sign-in pages or different service versions when trying to access Copilot.

• Disruptive Conversation Management: Features like the “stop recording” button in Teams can prematurely terminate interactions without recovery options.

3. Content Limitations and Overcautious Responses

• Strict Censorship on Certain Topics: Even neutral factual queries—such as election dates or legal processes—are often met with disclaimers or refusals.

• Lack of Actionable Responses: Copilot prioritizes general disclaimers over specific solutions, frustrating users who need more practical guidance.

4. Performance and Cost Concerns

• High Subscription Costs vs. Limited Features: At $30 per month, many users feel that Copilot’s capabilities do not justify its price, especially since it underperforms compared to ChatGPT and lacks some important features.

• Slower Performance: Compared to competitors like ChatGPT, Copilot is often sluggish due to its reliance on internal company data.

5. Privacy and Data Handling Issues

• Restrictive Data Access: Instead of allowing direct document uploads, Copilot requires integration with SharePoint, making it less useful for companies without a full Microsoft ecosystem.

• Unclear Data Usage Policies: Many organizations hesitate to fully adopt Copilot because they are uncertain about how Microsoft processes and analyzes sensitive business data.

Mitigation Strategies for Copilot’s Shortcomings

1. Improve Data Quality and Context

• Keep data sources up to date to ensure Copilot retrieves the most relevant information.

• Provide detailed context in prompts, including clear goals, specific data references, and necessary background details.

2. Enhance Prompt Engineering

• Use clear, structured instructions to guide Copilot toward more useful responses.

• Reframe prompts with positive instructions (e.g., "Summarize key action items" instead of "Do not miss any details").

• Iterate and refine prompts to improve accuracy over time.

3. Implement Review and Fact-Checking Processes

• Ask Copilot to explain its reasoning to uncover potential errors.

• Establish mandatory fact-checking protocols for critical reports and outputs.

4. Customize and Train Copilot for Your Needs

• Add custom glossary terms and synonyms to help Copilot understand company-specific terminology.

• Provide regular feedback to refine Copilot’s outputs and ensure it adapts to your workflows over time.

Recent conversations show that Microsoft Copilot can be transformative if implemented correctly. Some users value it for tasks like email summaries and meeting organization, while others—especially the already organized—view it as helpful but not essential.

The key to success lies in structured implementation. Companies that invest in initial training, pilot groups, and AI champions report higher adoption rates and better overall satisfaction. Regular training, active feedback collection, and managed rollouts help teams identify the best use cases and spread knowledge effectively.

I was a bit disappointed that I haven’t found anyone who is implementing or has already implemented Microsoft Agents. This could be a great differentiator for Microsoft AI products because, in theory, it provides an easy way to build AI-based automation seamlessly connected to all of their Microsoft apps.

If you are implementing or using Microsoft Agents or know someone who is, I’d like to chat about your experience.

AI Recommendations: Article, Video, and Tool of the Week

📄 Article: The Anthropic Economic Index – If you’re curious about how AI is actually being used in the workplace, this report is a must-read. Instead of speculation, it looks at millions of real-world AI interactions to understand where AI is making the biggest impact. The findings reveal a clear divide between AI-driven augmentation (enhancing human work) and automation (fully replacing tasks). It also highlights which industries are adopting AI the fastest and where it's struggling to make an impact. If you are wondering how AI might shape your job, take a look at this report.

🎥 Video: 10 Microsoft Copilot Tips & Tricks – If you're using Microsoft Copilot but feel like you’re not getting the most out of it, this video is a great place to start. While it is Microsoft-sponsored and a bit promotional, it still covers useful, real-world applications that can help streamline your workflow. The video walks through how to automate presentations, generate video summaries, set up smarter email rules, and use Copilot for quick data analysis. Even if you’re already familiar with Copilot, there’s a good chance you’ll pick up a new trick or two to make your workday smoother.

🛠 Tool: Whisk by Google – If you’ve been following my newsletter for a while, you might have noticed how my visuals have evolved over time. I started with DALL·E, then moved to Canva + DALL·E, and this week, I decided to try something new—Google’s Whisk. And I love it.

Whisk is Google’s latest AI experiment that lets you generate and remix images in a totally different way. Instead of typing out long, detailed prompts, you simply drag in an image for the subject, another for the background, and a third for the style. Then, you can tweak and remix them to create something completely unique.

If you're in the US, try it out today—I promise you’ll have fun!

Let me know what you think of these picks! 🚀

Closing Thoughts

As I wrap up this series on Microsoft Copilot, it becomes even more obvious that AI in the workplace is here to stay. However, how we implement it makes all the difference. Whether it’s Copilot, ChatGPT, or something entirely new, the key is structured adoption, training, and finding the right fit for your needs.

Now, I’m shifting my focus to building AI agents designed specifically for CFOs and finance teams. If you’re interested in shaping the future of AI in finance, I’d love your input.

🔗 Take the questionnaire here and help define what comes next.

As always, thanks for being part of this journey—exciting things ahead! 🚀

We Want Your Feedback!

This newsletter is for you, and we want to make it as valuable as possible. Please reply to this email with your questions, comments, or topics you'd like to see covered in future issues. Your input shapes our content!

Want to dive deeper into balanced AI adoption for your finance team? Or do you want to hire an AI-powered CFO? Book a consultation! 

Did you find this newsletter helpful? Forward it to a colleague who might benefit!

Until next Tuesday, keep balancing!

Anna Tiomina 
AI-Powered CFO