How the Paperclip Company Built Their AI Foundation with Microsoft 365 Copilot
Learn practical Copilot adoption techniques with business use cases through our story-based series.
Paperclip Co. has come a long way. What started as a cautious experiment with email summaries and draft replies has grown into something the team genuinely relies on every day.
Emma is surfacing operational insights she never had time to find. Liam is producing smarter, more targeted content. Sophie is resolving customer issues faster and spotting patterns that help prevent them.
Now Alex is back for one final session. Level 3 is the biggest step yet, and the most rewarding.
Where the Team is Now
Emma, the Office Manager
Emma is unrecognisable from the person who said “I’m not an analyst” just a few weeks ago. She’s been sharing meeting pattern reports with the management team and has already helped streamline two recurring meeting series. Level 3 brings a new challenge though: using AI to contribute to strategic decisions. Emma’s not sure that’s her place.
Liam, the Marketing Manager
Liam has gone all in. He’s segmenting content, tailoring campaigns, and has started experimenting with prompts on his own. If anything, Alex is slightly worried Liam might be a little too confident, moving fast and not always pausing to sense-check what Copilot produces.
Sophie, the Sales Assistant
Sophie has become the team’s quiet advocate for Copilot, showing colleagues small tricks she’s picked up. But Level 3 introduces something that gives her pause: using AI to generate product ideas and shape strategy. “That feels like it should come from people,” she says. “Not a machine.”
Alex’s Introduction
“Here we are, Level 3. I want to start by saying something I mean genuinely: you should all be proud of how far you’ve come. A lot of teams get stuck at Level 1. Some make it to Level 2. Not many get here. Level 3 is different from what we’ve done before. We’re not just using AI to work faster or smarter, we’re using it as a creative and strategic partner. That means bringing it into the conversations that shape the future of the business.
Today I want to introduce you to two capabilities you haven’t used yet: Copilot’s Researcher agent and Analyst agent. Don’t let those words intimidate you.
Think of Copilot Researcher as an incredibly thorough assistant who can dive into huge volumes of documents, emails, and reports across Microsoft 365 and pull out the insights that matter most. It summarises information, highlights trends, and surfaces key points so you don’t waste hours sifting through content.
Copilot Analyst, on the other hand, is like a data-savvy colleague sitting beside you. It can examine spreadsheets, databases, and business metrics, then translate that raw data into clear insights, trends, and actionable recommendations. Essentially, it helps you see the story your data is telling without needing to be a full-time analyst yourself.
I’ll be honest with you, this level will push you. And at least one of you is going to tell me this feels like it’s gone too far. That’s exactly the right reaction. Let’s see where it takes us.”
Level 3: Lead with AI Innovation
Emma the office manager’s lesson
Alex: Emma. Last session. How are you feeling?
Emma: Nervous, if I’m honest. I’ve loved what we’ve done so far but you said this level is about innovation and strategy. I keep thinking that’s not my job. I keep the place running.
Strategy is for the directors.
Alex: I hear that a lot. And I understand why people feel that way. But I want to gently challenge it. The people who keep the place running often have the clearest view of what’s
actually happening. They also don’t always have the tools to communicate it upwards.
Emma: That’s… actually quite true.
Alex: Let’s use that. You’ve been gathering meeting pattern data and workload insights since Level 2. That information has value beyond scheduling. Let’s use the Analyst agent to do something with it. Open your data, even just a summary you’ve saved, and try: “Using Analyst, look at the workload and meeting pattern data from the last quarter. Identify where the
business is losing productive time and suggest two operational improvements I could propose to leadership.
Emma: It’s identified that the business loses an estimated six hours per person per week to unproductive meetings, and it’s suggesting a meeting audit and a new internal communication protocol as proposals.
Alex: How does that feel?
Emma: Like something I could actually take to a director. With evidence.
Alex: That’s strategy, Emma. It doesn’t have to mean sitting in a boardroom. It means using what you know to make the business better. Try one more, let’s use Researcher to go a step further: “Using Researcher, find examples of how similar businesses have successfully reduced internal meeting overload. Summarise the top two approaches.”
Emma: It’s found two case studies. One company introduced ‘no meeting Wednesdays’, another moved to async updates for recurring status calls. Both reported significant productivity gains.
Alex: Now you have a proposal, internal evidence, and external benchmarks. You walked in here today thinking strategy wasn’t your job. What do you think now?
Emma: I think I might need a bigger notebook.
Liam the marketing manager’s lesson
Alex: Liam, I’ve been watching you work since Level 2 and you’ve been brilliant. But I want to ask you something directly: how often are you checking what Copilot produces before you use it?
Liam: I mean… I check it. Mostly. Sometimes it’s so good I just tweak a word or two and send.
Alex: That’s what I thought. And I get it, when something saves you this much time, it’s tempting to trust it completely. But Level 3 is where that habit can catch up with you, because we’re going to ask Copilot to do something much more ambitious: help you build brand strategy and generate new product ideas.
Liam: Okay, now you have my full attention.
Alex: Good. Let’s open Word and use the Researcher agent. This is a step up from what you’ve used before, it doesn’t just work with what you give it, it can pull in broader information to inform its output. Try this prompt: “Using Researcher, identify three emerging trends in sustainable consumer products that could be relevant to an eco-friendly stationery brand. For each trend, suggest one product or campaign idea Paperclip Co. could explore.”
Liam: This is… actually really impressive. It’s pulled in stuff about refillable product lines, carbon labelling on packaging, and a growing market for workplace sustainability certifications. And it’s given me a campaign concept for each one.
Alex: Now, which of those would you actually pursue?
Liam: Honestly? The carbon labelling one feels very us. The certification angle could work but it’s a longer play. The refillable line is interesting but we’d need to look at manufacturing costs.
Alex: That right there is the value of Level 3. Copilot gave you the raw material in minutes. But the judgement, like what fits your brand, what’s realistic, what excites you… That’s still entirely yours. Copilot doesn’t know Paperclip Co. the way you do.
Liam: So it’s less about trusting it and more about using it to think faster?
Alex: Exactly. It widens your thinking. You still decide what to do with it. And you still check everything, especially when the stakes are higher. A draft newsletter is one thing. A brand strategy is another.
Liam: Point taken.
Sophie the sales assistant’s lesson
Alex: Sophie, you’ve been one of the most enthusiastic advocates for Copilot on the team. But you mentioned before we started today that Level 3 made you uneasy. Tell me more about
that.
Sophie: It’s the idea of using AI to generate product ideas or shape strategy. That feels like it should come from people. From experience. From knowing your customers. If we just ask a
machine what to sell next, what’s the point of having a team?
Alex: That’s one of the most thoughtful things anyone has said to me in one of these sessions. And I want to give it a proper answer rather than just reassure you. You’re right that it should come from people. It always does. What Copilot does is help those people access more information, more quickly, so their decisions are better informed. Let me show you what I mean. Open Teams, go to a customer thread with a reasonable history, and open Copilot. Try the Analyst agent with this: “Using Analyst, review our recent customer interactions and identify the top three unmet needs or requests that come up repeatedly.”
Sophie: It’s flagging that customers keep asking about bulk order discounts, that several have requested eco-friendly packaging options, and that there’s a pattern of queries about delivery tracking.
Alex: Now, did a machine just decide your product strategy?
Sophie: No. It summarised what our customers have been telling us. We just hadn’t had time to join the dots.
Alex: Exactly. The ideas still come from your customers. The strategy still comes from your team. Copilot just means you don’t need to manually read through hundreds of conversations to find the signal in the noise. Try one more: “Based on recurring customer feedback, draft three product or service improvement suggestions I could present to the management team.”
Sophie: These are actually things I’ve been thinking about for months. I just never had the time to write them up properly.
Alex: And now you do. In about two minutes. That’s not replacing your judgement, Sophie, it’s giving it a platform.
Sophie: (quietly) Okay. I think I had the wrong picture of what this was.
What the team learned & what’s next
Emma, Liam, and Sophie came into these sessions at very different starting points. One was too busy to investigate AI. One thought it would lower the quality of his work. One hoped it might just speed up a few tasks. None of them could have predicted where they’d end up.
Here’s a final recap:
Emma (Office Manager) used Copilot’s Analyst agent to transform operational data into a strategic proposal for leadership, complete with internal evidence and external benchmarks. She came in thinking strategy wasn’t her job. She left with the tools to change that.
Liam (Marketing Manager) used Copilot’s Researcher agent to identify emerging market trends and generate brand-aligned product and campaign ideas in minutes, then learned the most important Level 3 lesson: that speed without judgement is a risk, not a superpower.
Sophie (Sales Assistant) used Copilot’s Analyst agent to surface unmet customer needs from conversation history and draft improvement proposals for the management team and in doing so, resolved her deepest concern: that AI was here to replace human thinking rather than support it.
Level 3 isn’t the finish line, it’s the point where AI stops being a tool you use and starts being a way you work. The team at Paperclip Co. have built their foundation, accelerated their engagement, and now they’re ready to lead.
So are you.