Last week, I spent an hour and a half with our claims management team exploring what Microsoft Copilot can actually do in a real work environment. Not the marketing promises or demo videos—actual, hands-on testing with their daily claims, emails, and documentation. What we discovered was both impressive and practical. I feel the best way for people to learn is by getting their hands dirty and playing around with what they are going to be learning with.

Summarizing Claims in Seconds

The first real test was claim summarization, and this is where the room started buzzing with excitement. An adjuster pulled up a complex multi-injury workers’ compensation claim they’d been managing—the kind with dozens of medical reports, multiple providers, and a tangled treatment history. They clicked the Copilot icon and simply asked it to summarize the claim.

The claims manager opened a complex multi-injury workers’ compensation claim in their browser—one with dozens of medical reports, multiple providers, and a tangled treatment history. They clicked the Copilot icon in the browser and asked:

"Summarize this claim"

Within about ten seconds, Copilot delivered a clean, organized overview with bullet points covering the key injuries, treatment providers, and current status. The room went quiet as everyone realized what they’d just witnessed—a task that normally takes twenty to thirty minutes had been completed in seconds.

But we didn’t stop there. We experimented with refining the prompt for different audiences:

"Create a brief, accessible summary of this claim for an insurance agent"

This produced a shorter, less technical version focusing on the essential facts.

Then someone asked:

"Provide a detailed summary including all compensable injuries and treatment timelines for reinsurer review"

This time, Copilot generated a much more comprehensive report with dates, specific medical terminology, and treatment progression—exactly what would be needed for formal review.

Finally, we tested an executive-level prompt:

"Create a concise executive summary highlighting key concerns and next steps"

The output was perfect for leadership: high-level overview, critical issues flagged, and clear action items.

Each request produced appropriately formatted output within seconds. The difference in quality between “Summarize this claim” and the more specific prompts was striking—proof that the tool responds to the level of detail and context you provide.

Within about ten seconds, Copilot delivered a clean, organized overview. What would normally take twenty to thirty minutes of reading through notes, cross-referencing dates, and writing up a coherent summary just… appeared. But we didn’t stop there.

We experimented with tailoring summaries for different audiences. When preparing information for an insurance agent, the claims manager asked for a brief, accessible summary. For a reinsurer review, they requested detailed information including all compensable injuries and treatment timelines. For executive leadership, they asked for a concise summary highlighting key concerns and next steps. Each request produced appropriately formatted output within seconds.

This wasn’t some cherry-picked demo claim either. We tested it on several different files, including ones that had just been transferred to new adjusters who weren’t familiar with the case history. The tool consistently delivered useful summaries that captured the essential information.

Taming Email Chaos

If there’s one thing every claims adjuster deals with, it’s email overload. Defense attorneys, medical providers, claimants, internal teams—hundreds of messages daily, many of which contain critical information buried in long threads or scattered across weeks of correspondence.

We tested Copilot’s ability to search and synthesize email content, and honestly, this might have been the most valuable discovery of the session. One claims manager asked Copilot to summarize all emails related to a specific claim number. Within seconds, it had searched through months of correspondence and provided a chronological overview of key communications.

But the really impressive moment came when someone asked about compensable body parts for a claim they’d recently inherited. Copilot didn’t just search email subject lines—it actually read through the content, found a large loss preview document that had been attached to an email weeks earlier, and extracted the specific information: fractured pelvis, rotator cuff tear, neck injury, and hip injury. Information that would have required opening multiple emails, downloading attachments, and reading through pages of documentation was delivered in seconds.

What made this even more powerful was the conversational follow-up capability. After getting an initial summary, adjusters could ask

"Was this issue resolved?"

or

"What was the outcome of that defense coordination?"

and Copilot would search through subsequent emails to provide answers. It felt less like using a search tool and more like having a conversation with someone who had perfect recall of every email ever sent.

The reaction to the email discovery was visceral. When Copilot didn’t just search email subject lines but actually read through message content, found that large loss preview document attached to an email from weeks earlier, and extracted the specific compensable body parts—fractured pelvis, rotator cuff tear, neck injury, and hip injury—you could see the lightbulbs going off around the room.

One claims manager literally sat back in their chair and said something like “Wait, it actually read through the attachment?” The realization that this wasn’t just keyword searching but actual comprehension of document content was the turning point. The managers started talking at once about their own inherited claims where they’d spent hours digging through email histories trying to piece together case details.

The most telling reaction came when someone said “So I don’t have to open twenty emails and three attachments just to figure out what body parts are accepted?” The frustration in that question—followed by the relief when the answer was yes—captured what made this feature so valuable. This wasn’t about saving a few minutes. It was about eliminating one of the most tedious and error-prone parts of claims management.

What really amplified the excitement was the conversational follow-up capability. When we demonstrated asking “Was this issue resolved?” and having Copilot search through subsequent emails to provide context and outcomes, people started leaning forward in their seats. One adjuster commented that it felt like having a colleague who had perfect memory of every email ever sent about a claim and could answer questions instantly.

There was also a moment of slightly nervous laughter when someone realized just how much information Copilot could access. “It’s scary, people,” one team member joked, which led to a quick reinforcement of the security protocols and the importance of using the Work tab. But underneath the humor was genuine appreciation—and maybe a hint of relief—that such a powerful search and synthesis tool existed for managing their overwhelming email volumes.

Medical Records Analysis

Medical documentation is probably the most time-consuming part of claims management. Adjusters regularly need to review hundreds of pages of medical notes, treatment records, and bills to understand what’s compensable, identify treatment patterns, and prepare reports.

We demonstrated how to download medical records as PDFs and upload them to Copilot for analysis. The questions you could ask were surprisingly specific.

"What body parts were accepted as compensable in these records?"

"Summarize the treatment history and timeline."

"List all medical specialists this patient has seen."

"Are there any gaps in treatment that should be noted?"

One team member had an immediate realization: “So I can take all the medical notes attached to bills in our system, save them as one PDF, and just ask Copilot to summarize everything?” Yes. Exactly that! What would normally be an hour-long review session becomes a five-minute exercise in asking targeted questions and reviewing synthesized answers.

This doesn’t eliminate the need for adjuster expertise—you still need to know what questions to ask and how to interpret the answers. But it dramatically reduces the time spent hunting through documents for specific information.

Throughout the training, I emphasized a key concept that resonated with the team: think of Copilot as your “second brain.” This framing helped claims managers understand how to integrate the tool into their thinking process.

Claims adjusters constantly juggle information—dozens of open claims, each with medical histories, legal complexities, treatment timelines, and correspondence threads. The cognitive load is enormous. Copilot functions as an extension of your memory and analytical capacity, augmenting your ability to access and process information without replacing your judgment or expertise.

When you inherit a new claim, your second brain instantly recalls every email, document, and note. When writing a summary, it has perfect recall of details you’d otherwise hunt down across multiple systems. This framing helped the team understand why prompt specificity matters—you ask Copilot targeted questions just like you would a colleague, and it responds with the precision your instructions deserve.

One claims manager said it felt like “having a colleague who had perfect memory of every email ever sent about a claim.” That’s the right mental model. Your second brain doesn’t get overwhelmed, doesn’t forget details, and doesn’t need to search through dozens of files. It’s always ready to surface the information you need instantly, freeing your mental bandwidth for complex judgment calls that require human expertise.

The Art of the Prompt

About halfway through the session, they discovered the single most important principle for using Copilot effectively: specificity matters enormously. One claims manager asked a vague question

"Tell me about this claim"

and got a generic, barely useful response. We refined the prompt to something much more specific:

"Summarize the compensable injuries, current treatment status, outstanding medical bills, and next required actions for this workers' compensation claim"

The difference in output quality was dramatic.

Someone on the team summed it up perfectly: “It’s all about how you word things. If you’re very vague with it, it’ll be very vague. If you’re very specific with it, it’ll be very specific.” This became our working principle for the rest of the session. Copilot isn’t magic—it’s a tool that reflects the quality of your instructions back to you.

We started treating it like briefing a highly capable but very literal assistant or a new hire. The more context and specificity you provide, the better the results. This actually made the tool more accessible, not less. Once people understood they needed to be clear and specific, they got better results almost immediately.

Data and Reporting

We spent some time at the end of the meeting exploring Copilot’s capabilities with Excel spreadsheets, particularly for filtering and analyzing claim data. This area presented more of a learning curve—there’s a difference between using browser-based Copilot and opening it within Excel itself, and we had to work through when to use which approach.

But the fundamental capability is there. Being able to ask

"Show me all medical-only claims from this spreadsheet"

or

"Filter this exposure report for claims exceeding a certain threshold"

Has clear value for reporting and analysis. We’re still refining best practices here, but the team could see the potential for streamlining regular reporting tasks but that is for another meeting.

What Actually Works

After testing Copilot across multiple use cases, here’s what we learned about its real-world capabilities. It excels at tasks involving synthesis and summarization—taking large amounts of information and distilling it into something coherent and actionable. It’s exceptional at searching across multiple sources simultaneously, whether that’s emails, documents, or browser tabs. And it handles conversational follow-up naturally, allowing you to progressively refine your questions without starting over each time.

The time savings are substantial. A claim summary that takes thirty minutes to write manually takes two to three minutes with Copilot. Searching through email history for claim details drops from fifteen minutes to under two minutes. Extracting compensable injuries from medical records goes from an hour of reading to ten minutes of targeted questioning. For an adjuster managing forty to sixty claims, these minutes add up to hours every week.

But Copilot isn’t a replacement for expertise. It works best with information that’s already documented and accessible. It can synthesize and summarize, but it can’t make judgment calls that require deep claims knowledge. It’s a tool for eliminating administrative burden, not for making complex decisions about case strategy or claim handling.

The Learning Curve

By the end of the session, the team was simultaneously excited and mentally exhausted. “We know just enough to be dangerous now,” someone joked, and that was exactly right. Most adjusters were producing useful results within thirty minutes of hands-on practice, but mastery clearly requires daily use and experimentation.

The team understood that this isn’t about perfection on day one. It’s about incorporating the tool into daily workflows and learning through trial and error what kinds of questions work best for different situations. The technology requires some finesse, but the payoff for developing that skill is significant.

The Real Value

As we wrapped up, one team member noted that this tool helps ensure their work remains manageable and sustainable. That really captures what we saw. For claims professionals managing overwhelming volumes of documentation, email, and reporting requirements, Copilot offers something genuinely valuable: time.

Time to think more strategically about complex cases. Time to communicate more thoroughly with claimants and providers. Time to make better decisions instead of just processing information faster. The tool doesn’t replace the adjuster’s expertise—it removes the obstacles that prevent them from applying that expertise effectively.

The session validated something important about AI in professional work environments: when implemented thoughtfully, it’s not about disruption or replacement. It’s about augmentation. These claims adjusters left the training excited about having a tool that could handle the tedious parts of their work so they could focus on what actually requires human judgment and expertise.

That’s what good technology should do—make competent professionals even better at their jobs by eliminating unnecessary friction and freeing up mental bandwidth for the work that truly matters.

---

If you’re planning to implement Copilot training in your organization, start with the highest-volume, lowest-complexity tasks like email summarization and basic claim summaries. Let people build confidence and proficiency there before tackling more complex applications. The wins will be immediate, and the enthusiasm will build naturally.