Transcript for: The AI-Powered Proposal Webinar
Jason Anderson:
With that, let’s just jump in. We’re a minute or two past the hour, and as people join, they can catch up quickly. I hate to make everyone wait.
Welcome, everyone—I’m Jason Anderson. I thought I’d share my video up front since I’m not sure how it looks in Zoom once I start presenting from my screen. So, hello—face to face. I’m excited to share this conversation.
We’ve got a great mix of folks here today: some current customers, some who’ve had a few discussions with us and our sales team, and others who are brand new. Welcome, and I’m looking forward to chatting.
Today’s topic is a fun one, I hope: the new world we’re living in with large language models and AI—how AI is evolving, and how we can go to work as proposal professionals. I want to talk about how we coexist with AI, how our software works, how AI works, how they work together, and offer some practical applications of what you can do with AI today.
As I mentioned, I’m Jason. I’m the head of sales and marketing here at Expedience. I’ve been here about seven or eight years. I came to Expedience as a customer—I had bought our software—so I bring the perspective of both having been a customer who knew I needed a more efficient way of creating proposals, and also the experience it took to deploy the software and get value from it.
I’m a longtime proposal writer and consultant—a champion of proposals—and a technology advocate as well. That’s a little bit about me.
Before I dive in, a quick Zoom overview: you’ll see the Q&A and chat panel at the bottom of your controls. Feel free to drop questions as they come up. I also have a colleague on Zoom right now who will do his best to respond to questions in real time. Toward the end, I’ll take a few minutes to share answers with everyone.
My contact details are beneath my photo—feel free to reach out if you want to chat further, if we don’t get to your question, or if you’d like a deeper or different explanation.
What we’ll cover today
We’ll talk about AI in proposals generally, including:
- What it means from a security perspective
- What it means for individual accountability and “agency” (who owns the final content)
- Proposal automation and content libraries versus AI—where they’re similar, where they differ, and how they can be synergistic
- Prompting—some prompt examples
- A tool of ours: a prompt library, which helps centralize and reuse your best prompts
Quick overview of Expedience
For those newer to Expedience: we’ve been automating proposals since 2014 (that was the first version of our software), though the company was founded earlier. Our vision is to bring proposal automation and efficiency into the applications customers already use to create proposals—especially Word.
Many tools in the market try to move you away from Word into an online platform, then export back to Word. That approach may work for some organizations, but our approach is different: we bring those tools directly into the applications customers frequently use.
As Word has evolved, this strategy has paid dividends for us and for customers. Looking back over the last decade: Word introduced co-authoring (collaboration in the same document), notifications and @mentions, and writing-assistant capabilities. More recently, Microsoft released early components of Copilot and generative AI, which is what we’ll focus on today.
Copilot as an AI hub inside Microsoft 365
Copilot has become a powerful hub for AI productivity within the Microsoft Office suite. Because it lives inside Microsoft 365, it provides accessibility (it’s where you already work) and also security. It fits into the security landscape your organization already has in place for Microsoft 365, so there’s no incremental exposure just from using Copilot. I’ll talk more about that in a moment.
“Is it safe?”—Security and privacy
A common first question is whether Copilot is safe. In my conversations with customers, I highlight that Copilot respects permissions: it can only access what the user can access. It also operates within Microsoft 365’s security and compliance framework. And importantly, there isn’t an opportunity for your data to be retained and used for training in the way people worry about when employees paste sensitive content into consumer tools.
A common myth: “Can AI just write my proposal?”
Customers often ask: “Can AI just write my proposal? I have an RFP—can I have AI look at old proposals and create the best winning response out of the box?” The short answer is: it might look like it can, but the real answer is no—not in a trustworthy, hands-off way.
Why? Because you have two challenges:
- Your content (old proposals) can be outdated, inaccurate, or inconsistent. If you’ve ever searched SharePoint for old proposals and copied content forward, you know how risky that can be. Copilot can do it faster—but it can still pull forward the same underlying issues.
- AI itself can be a problem. AI can combine accurate things into an inaccurate conclusion, or fill in gaps with invented information. That’s what people call hallucinations. So you must read and proof every word that comes out of AI—including Copilot.
Agency and accountability
This is important: when you sign off on a proposal and submit it—whether internally or to a customer—you own it. That’s why trustworthiness matters. AI can help a lot, but you need to know what you can count on and what must be verified.
Where proposal automation fits
So where does Expedience—or proposal automation—fit? The bottom line is that it’s about building a trusted source of content so you can create documents more efficiently without re-proofing everything from scratch each time.
You do that by taking old proposals and curating them: selecting the best, most current, most accurate content, sanitizing it, removing customer-specific details, and rewriting it so it becomes reusable. Those “Lego blocks” become assets in the Expedience content library.
Not everything can live in the content library, though. There will still be times you need to pull something from old content, scrub it, improve it, and then—ideally—turn that “nugget of gold” into a vetted content-library asset for future reuse.
How Copilot fits with proposal automation
A great use of Copilot is finding content (“Where did we write that before?”), reimagining it for a new audience or industry, and helping unify voice across multiple contributors. Copilot is also excellent at helping you meet character limits (for example, in Excel-based RFPs), condensing content, analyzing RFP requirements, and grading your responses against requirements—again, with the reminder that you must proofread.
Demonstration examples (conceptual walkthrough)
Expedience lives in Word as a set of ribbons. The content library is presented as categorized dropdowns and galleries—reusable elements you want to keep consistent and accurate (resumes, legal terms, process descriptions, and so on).
Example 1: Answering a questionnaire using the content library
If a customer asks, “Describe your company’s history and vision and culture,” you can highlight that question and search the content library to find the most relevant vetted content, preview it, and drop it into the document. The content is structured for personalization without accidentally carrying over another customer’s name.
Example 2: Template-driven proposal creation (PDQs)
We also have template-driven documents that behave like a wizard: you select product lines, service areas, pricing models, and other options, and the proposal assembles using trusted content-library records. This can look visually similar to AI because content “appears” quickly—but the difference is that this output is rooted in vetted content.
Prompting with a prompt library
To make prompting faster and more consistent, I also showed a prompt library tool that helps centralize and reuse best prompts. For example, when reviewing an RFP, you might:
- Provide Copilot context about your company (history, strengths, differentiators)
- Ask Copilot to do bid/no-bid analysis
- Ask it to summarize the RFP for SMEs
- Ask it to propose win themes, risks, gaps, and desired outcomes
Copilot can produce a helpful starting framework—but it still must be scrutinized and verified.
Rewriting technical SME content
Another example: you might receive technically accurate content from an engineer that’s heavy with jargon and acronyms. Copilot can rewrite it to be more understandable and persuasive for a broader audience—defining acronyms, removing jargon, and adopting a professional tone. In many workflows, you’d still send that rewritten version back to the SME for accuracy sign-off.
Wrap-up and Q&A
Those are the core examples I wanted to share. There are many more, including summarizing not just across a single document but across multiple documents—like a SharePoint folder of recent proposals—especially when you scope the source carefully.
Jason:
Do we have any questions you’d like me to address now?
AJ:
Yes, I was able to handle them throughout.
Jason:
Great, wonderful.
One last point: Copilot can pull from old proposals, but you can also point it at your vetted, curated content library as a trusted source. These approaches can be highly synergistic.
I know that’s a lot to cover in 40 minutes, but I hope it was helpful. I’m happy to dig deeper. My contact details are on screen: Jason@exp-soft.com. AJ is also here, and many of you have spoken with him. Feel free to reach out to either of us or anyone else on the team. Thanks so much.