Jasper talks about validating the import of your most complex source documents and portals. Validate the branded export. Only then judge the workflow in between. A tool that drafts well but fails at either boundary is, in his words, manual-to-end or end-to-manual, and you will be doing the first or last hour of every bid by hand.
From there the conversation moves through the parts of the process AI has already absorbed (go/no-go analysis, first drafts, consistency checks), why teams keep using SMEs as a safety blanket they no longer need, and the volume and FTE-hour thresholds that tell you when a platform earns its cost. Jasper and Christina also get into the build-it-yourself question (Claude, MCP connectors, Copilot) and where that route carries real compliance risk, how to answer a CISO who has banned AI in the sales process, and what 2027 does to win rates when every team automates at once.
Jasper Cooper spent more than a decade running enterprise sales and RFP processes before building AutoRFP.ai, and has since worked with over 200 companies moving from legacy to AI-first proposal operations.
See AutoRFP.ai and the full end-to-end category in the 2026 Proposal & Bid Software Report: https://stargazy.io/offers/2026-proposal-and-bid-software-report
THE STARGAZY BRIEF ✹ EPISODE 25
Guest: Jasper Cooper, Co-founder & CEO, AutoRFP.ai Host: Christina Carter, founder, stargazy Topic: End-to-end proposal management software ✹ how to evaluate it, when it makes sense, and where the category is heading
Christina Carter (00:08) I have spent the last year evaluating well over 40 vendors And in this, I am talking to Jasper Cooper, who's the co-founder and CEO of AutoRFP. Jasper spent over a decade running enterprise sales and RFP processes before building our AutoRFP, and has since worked with over 200 companies moving from legacy to AI first proposal operation. So...
If you're trying to figure out which platform to buy and why the last one didn't fix what it was supposed to, stick with us and you're going to figure out if this is a category that you should be paying attention to. All right, into the episode.
Christina Carter (00:39) Hey Jasper, welcome back to the Stargazee Brief, especially today talking about, you know, proposals and the very, very different type of proposal management systems out there. Just because I do want to help people kind of understand like where each one fits and maybe which one is best for them so that if they are thinking about the tools they're using or the tools they want to use, they can kind of figure out what that looks like. So thank you so much for hopping on and kind of helping us out here.
Jasper Cooper (AutoRFP.ai) (01:04) Yeah, always glad to be invited back.
Christina Carter (01:06) the last time we talked, we talked about knowledge management, in proposals specifically, and you're actually my first guest on the Stargazee Brief. But, like this time, I really do want to focus on a proposal management category that I see as like end to end proposal management.
I think the difficult though part Jasper is almost every single vendor out there says that they are end to end proposal management software in some way. So let's say you're a head of proposals, head of bids, maybe you're a head of revenue and you're trying to buy a new proposal management software. You know, you need an end to end proposal management software. How can you tell the difference between that and a soft or proposal software that isn't end to end? Like what questions should I be asking?
Jasper Cooper (AutoRFP.ai) (01:50) Yeah, I think a really interesting one, it's because of the technical difficulty on the vendor side is just looking at the very start and the very end of that process. That's not very AI, very old school, but it's like very hard engineering work, which is one being able to actually import the documents, the portals, whatever you're primarily working with, maybe even the most complex examples and get those into the system cleanly in a way that makes sense. So validating that and then.
Also on the very back end of it, validating that it can export in the way that you need, right? Do you need it branded in a nice proposal template? Does it support all of that? Does it support going back into the customer document and retaining their formatting if you need that? But I think those are really like easy weak points to discover there. And then of course you have the whole kind of middle, right? After you get it in the workflows and such, the reporting. And I think, yeah, different vendors will have different.
capabilities there, but I guess just testing the start and the end is like a really important part that some people work around and then you realize it's not necessarily ends to end. It's like manual to end or ends to manual. Yeah.
Christina Carter (02:55) Yeah. No, I think that's, that kind of goes to my next question, which is, I mean, the phrase everyone uses right now is human in the loop, of course. So what would you say a human is really doing right now? Like, let me step back, rephrase that question. So like, if I, if I am a proposal manager, like I manage a proposal team in some way,
What should I be expecting proposal people to need to do right now within that process? And what should I be expecting a software to be able to do to support, you know, the person who's actually doing the work?
Jasper Cooper (AutoRFP.ai) (03:30) Yeah, like I think we're past the point of first drafts now, right? Everyone should have some sort of first draft and that should be in there. The relevant content should be somewhere easily accessible so we can do that. Like they shouldn't be starting from scratch where there's answers that exist, right? So that's the basics. shouldn't, bid writer's job or a bid manager's job shouldn't be to get in there and actually start copy pasting like fresh responses and it should be able to do the absolute essentials.
And then from there, also trying to find like the most relevant subject matter experts for those particular types of questions. That should be easy, but I think we still have very much a human in the loop. They're trying to find the right individuals at the org and assign it and kind of dealing with a lot of that project management aspect at the moment. I think that's a, that's a fair thing for them to be involved in. And yeah, I think it's interesting right now because.
Although you have the ability to pull through content, it's still questionable for most organizations if that content is actually accurate and up to date. And that's kind of breaking the current frontier because you still have to get in there and really at least quickly review everything. But for some teams, they're starting to get to the point where they have trusted sources for certain topics and they can approve and they can actually remove.
a human or a subject matter expert in that case from the loop, speeding it up. So I think that's like one of the expectations that's going to change really, really significantly is like now this team doesn't have to get involved or be that human in the loop. And now for these types of questions that aren't that sensitive, they can be outside of that. And I think we're going to see a lot of both companies and tools just like focus on working together to closing the loop and closing more and more humans out of that loop that definitely don't want to be involved in part of
Christina Carter (05:13) Yeah, yeah, I think that's a really big part of it is I think you can like right now I think especially in like mid-sized to large organizations you usually have like 10 to 12 people for every single RP who are helping respond and most of those people aren't they don't work in bids like that's not their job. So yeah, I think any way you can help them is is really important. What you seeing are you seeing like specific process pieces?
Jasper Cooper (AutoRFP.ai) (05:49) Yeah, yeah. think like go-no-go analysis is like the easiest to apply thing, right? That large language models were just perfect for out of the box, just taking all the responses and doing the go-no-go analysis you need. So I think that's mostly being handled as well. So that kind of started the intake process, the drafting we've spoken about. And I think the immediate next thing that we'll see roll across all systems exists in some, maturing in others is automatic assignment on top of that.
So we'll see that more accurate. Previously, it was like a key word. You know, if it says security, then get the security person, but we need to move past that to make sure we're not using them as a, as a safety net for no reason. Right. We're only pulling them in where the system isn't confident that that's a hundred percent the correct response. And we actually need their help right now. I feel like we're using a lot of subject matter experts as like a warm blanket, like, please check this. Like, even though we don't need to, need to involve them. ⁓ so I think that that aspect of it.
And then like post analysis, there's different types of post analysis that people do, but I think a lot of those post analysis have been automated, you know, from review this entire document and make sure there's no inconsistencies in facts. You know, that's a very easy kind of thing to do. So less and less time being spent there. And I think now it's into the more nitty gritty stuff that will be really an engineering challenge. I think for all vendors as they start to apply this technology to the project management, which has always been.
No, I know you've been a big proponent for that as far as there needs to be more automation, more focus on functionality when it comes to managing and hurting all the cats. So I think that's really next is the easy kind of low hanging fruit. A lot of them have been picked or being picked now and will start to kind of get higher up the tree. And I think I think the top of the tree there is really autonomous project management, some sort of ambient system that is able to follow up with sneeze where that makes sense for you automatically as well.
Christina Carter (07:42) Yeah. Can we make that a thing? cause that's literally everybody's least least favorite part. Well, I think most people's least favorite part of the job is, is doing that, that piece. And then, I guess like, obviously an end to end proposal management system is, is it's a lot, which is good for most people. It's good for most teams, for a lot of companies, but not every single time and not for every single company. So.
Christina Carter (07:42) I'm curious what your opinion is. Like when does an end to end proposal management system make sense? Like for what team size, industries, things like that. And when does it not make sense? Like when should they be looking for something else? Like whatever that might be.
Jasper Cooper (AutoRFP.ai) (08:18) Yeah, there's different ways to look at it. I think one of the easiest, like clean ones, just volume, right? If you're doing one RFP a month and they're not like huge, you know, mega projects with many, many people involved and you are just doing that one. Um, and I think, yeah, you're definitely in chat GPT Claude territory for whipping that up and getting that out in the world. Um, you know, use a, use a spreadsheet to track the project management, et cetera. When you get into like two.
And like, of course you can, you know, if you can buy nice tools, then great. That's awesome for you. But I wouldn't say that's like mandatory. then, you know, when you get into two RFPs per month and they're pretty consistent. That's pretty good idea. That makes sense. Particularly if they're more complex or large or you're involving, yeah, like eight people, you know, in the process throughout then collaboration aspect gets a little jarring. then I think one per week, 50 per year is when you're starting to get into like, yeah, this actually makes sense. Most people at that scale do have a tool, depending on what sector you're in. ⁓ And then hundreds plus, still see people in the hundreds plus RFPs range, multiple people without tools. I think that's borderline insanity at that point.
They're higher and higher and higher and higher and don't really work on the manufacturing line, right? They're still doing it all very manually, the old school way, the artisan kind of way. So yeah, hate to see that. I think there's another way to look through the lens as well is it's not, maybe it's not just volume, but another metric you could use that would be useful is like the amount of full-time equivalent FTE hours you're spending on it. I think if you've got a dedicated bid manager.
It's really good to pair that with a tool so you're getting some leverage so you don't need to immediately move to two and then three and then fours as things scale you can get some really good efficiency get a win rate up as well with like an incremental investment and maybe you're like coming from the software side solutions engineers they're cumulatively across five of them spending a full-time equivalent on it I think that makes sense to leverage a tool as well there but yeah interested your thoughts on that like does that feel about right?
Christina Carter (10:32) That does feel about right. I genuinely think if you have one person who is having to set aside their full-time job, if they're not a proposal person, to respond to an RFP at least once a week, then yeah, you need a tool to help them out. If not only for their sanity, but for, like you were saying, not just even time-saving, it's actually, I think, helps reduce risk.
Like it's so easy to make mistakes with these RRPs, both in public and private sector. You know, this kind of like helps you make fewer mistakes. It makes you don't miss something. So you're compliant when you submit something, you don't have procurement coming back and saying, ooh, you're out because you didn't submit whatever this is, or you submitted something incorrectly. I think it just like helps you do that, right? So to me, that's where it's really important. Something I had to think a lot about and I have an opinion about it, but is specifically if you are a business and you don't have a proposal manager and you're not going to hire a proposal manager for whatever reason, like maybe you're in tech and you have a solution consultant, maybe you're an AEC and you just have SMEs, like your engineers do it. Like, do you think they need an end to end tool or do you think they just kind of either like a mash of like cloud skills and plugins or should they be using like an automated response tool and like, what are your thoughts on this?
Jasper Cooper (AutoRFP.ai) (11:56) Yeah, I think that that's definitely more nuanced, particularly when you go into AEC, like the proposals there are just so, the complexity of them ranges so much that I think it becomes a question of if there even is a tool that would give you meaningful efficiency, you know, that exists on the market. So for that kind of setup. So I think you'd need to evaluate it it becomes like a genuine question. Like, yeah, will this save me? Like if I'm including all of my SMEs and then they're heavily customizing it every time, and it's very engineering focused, and it comes from math, and it comes from this, and it's not really quite as simple as tailoring previous responses for a majority of the response, then it's going to be really hard to get leverage out of AI, right? So you're looking more at collaboration and process and how much you're spending there. So you could start to bucket your time to try and figure that out. So I would say there, yeah, definitely not always. And that would come back to volume.
Maybe if you're spending six hours on the coordination side, then it only makes sense once you're getting to coordinating three projects. Even though you're technically spending 100 hours on bids, a lot of it isn't easily automatable because you need to build schematics. And that's just not going to be automated for you today, at least by our proposal software. Yeah.
Christina Carter (13:11) Yeah, that makes sense. mean, but what about like for proposal, like, sorry, what about like software? I like, let's say you're in software, you have a team of solution consultants, you get RFPs in, you have them respond. Would you suggest that they need an end to end proposal management tool if they are responding to enough? Or do think that's like almost too complicated and that they should just be using something a little bit more simple?
Jasper Cooper (AutoRFP.ai) (13:34) Yeah, it is interesting with the simplicity side. think if you've got a solutions engineering team and there's like multiple people and you're receiving a reasonable volume of RFPs, like you're a pretty big organization. You've got pretty big and complex products. And you also have a solutions engineering team, not just to do RFPs, right? But primarily to do demos and account setups and proof of concepts and so on and so forth. And you're receiving them regularly enough for this to be a pain point. I would say definitely.
It's worth investigating a platform at that stage because those people are going to get leverage off each other. The hard part won't be necessarily learning an RFP software as annoying as that is. It will just be the compounding gains of everyone doing it in the same tool and being able to share those answers. Because at the end of the day, we can plug in some Claude MCPs and get the documentation and things in there, but there's still going to be gaps in the knowledge. It's not going to be necessarily appropriate to like go and build a new skill or build a new doc.
Add that every time and then you're just starting to do things manually that are very old school in a way, just with kind of the Claude top hat on it. Like I'm updating the library in Excel that I use in Claude. Like, yeah, that's great and all, but it doesn't give you any like real sense of efficiency. And if you're in a compliance heavy space, then yeah, there's like real risk to that approach as well.
Christina Carter (14:47) Yeah, I agree. mean, like, I think if you are receiving a certain amount of RPS, you should probably just hire a professional to do it, along with getting an end to end bid management software. That's my that's just my opinion. And like, and we see we see revenue goes go up with that. So that's my suggestion. But but I mean, like, it's even if you didn't like you, you do need, I think, software. I do think, like, for the reasons why I was saying kind of before, like, I do think it does reduce risk.
I think you're more likely, like obviously you want speed, your solution engineer doesn't want to be up every single night and spend their weekends responding to an RP because during their day they don't have time to respond to the RP. Cause that's probably not in their job description. So yeah, it's to me, it's almost like a no brainer to get an end end software, but that's just me. So I wanted to also talk about
Christina Carter (15:12) something that maybe you won't love talking about. think probably a lot of proposal, not just proposal, but like any software CEOs don't love talking about, like we have Microsoft co-pilot, it's already in SharePoint and Word. Google has gems and it has Gemini and Workspace and Cloud Code can do so much, like really cool automation, like genuinely cool automation. so a lot of teams have like literally come to ask me, Hey, just, can you just build out?
Jasper Cooper (AutoRFP.ai) (16:12) Yeah, I really like like I'm a super heavy Claude coach, HPT user, et cetera, et And even for like one page proactive proposals, a lot of the time I'll use Claude. Like it's just fast and easy to get that simple stuff out. And I'm someone at the company who like, don't, one thing about like proactive proposals is they're generally pretty low stakes, like they're for mid-market or small businesses.
And then two, I have all of the knowledge required to just skim something and be like, that's true and correct. Like I'm not worried about any, any risk or it being out of date in a really simple doc. So I think there's absolutely cases where you should just use cord, some MCP connectors, Google drive, figure that, figure that out. and it's a very fast way to do what you might be doing manually right now. You might be able to one shot it with a, with a nicely put together cord skill. on the other side, I think like.
What I've started to explain, it's interesting with different people, like at different tiers that ask this question, but when I get this question, like the base level, it's like, if you come to ask me, you know, I'm building a tree house, I'll be like, yeah, you should use a, you should use a hammer for that. You know, just they're inexpensive. They're great. They work. There's all these different applications for them. You buy them for $12, whatever that looks like. But if you plan on building houses for a living, probably would go with a nail gun.
It's going to be more expensive, that's for sure. You're going to need to learn how to use it as well. But within a few minutes, I think you'll understand why I recommended it. And then every nail is going to go in faster, it's going to go in straight, it's going to be safer to use at that kind of scale. there's that. And also, if you're a professional house builder, you don't want to be the person holding the hammer when all the competitors have the nail gun.
Christina Carter (17:40) Yeah, of course.
Jasper Cooper (AutoRFP.ai) (17:56) It's not going to be a great time. that's kind of the lensing and framing that I use to kind of explain that at a super high level. And then of course, like there's so many good case by case details of just comparing those workflows side by side when you're using them for actual RFP volume and like actual complex bits.
Christina Carter (18:14) Yeah, I, I agree. just, I did a workshop yesterday with people who, have never really, most of them have never used quad. Like they're not allowed to use quad. but I was showing them how to build quad skills, like to automate pieces of their proposal software.
But the vast majority of them are not allowed to use AI what do you say to those like..CSOs or, you know, whoever's making that decision within the company to say like, look, like you should be using AI, in some way within your proposal responses or your proposal creation.
Jasper Cooper (AutoRFP.ai) (18:44) Yeah, yeah. And you mean like on the chief security officer side particularly, like chief information security and such?
Christina Carter (18:48) Yeah, well, yeah, because I have no doubt that you're having these conversations, like people pushing back saying like, we don't feel comfortable using AI within any of our sales process.
Jasper Cooper (AutoRFP.ai) (18:53) yeah.
A hundred percent. And I think a lot of these people were actually like burned by, the model companies early, like when you use a Gemini out of the box, like it's just training on your data. And then like, even if you go and you're an application provider and you use certain APIs from them, turns out they're training on your data. Like they're actually sending it to data centers in Kazakhstan. Like it's insane. so I definitely understand where the, where the, the genuine fear comes from.
And it's interesting when you try and put in a tool like that, because is, are you going to get on the phone and be able to have a detailed conversation with someone at like Anthropic or Gemini or something like that? So I do think you need to get a really good level of resolution on a tool, unfortunately, and understand where it's hosted, how it's trained, and if you can get that information to that stakeholder.
and even get someone from that organization who can get on a call to speak through the specifics and why it's not training and how it's not training. Unfortunately, I don't think it's just as simple as like going to them and saying, Hey, you know, this is secure. Everyone's using it. Can I please use it? You know, like I wish it was that that simple in a way, but there's actually real reasons why they're super concerned about it. particularly when, you know, in our space, there's a lot of AI startups.
Christina Carter (20:02) It's fine.
Jasper Cooper (AutoRFP.ai) (20:15) So it is this genuine risk there that certain startups aren't doing it the right way or whatever. They've probably burned by that in their career as well. So you just really want to try and do as much of the due diligence for them, try and understand some of those detailed concepts. Maybe your personal Gemini can help you understand some of those concepts and help you put that kind of due diligence together, send that questionnaire to the vendor, get them to fill it out and really come armed with things like zero data retention is a great one, right? So they're not.
saving your data, zero training is important. Also where they're hosting the data, where the storage and processing is. And if you're based in Europe, then there's probably some European requirements. you're in Switzerland, it gets potentially even crazier than that on the financial services side. So know those types of requirements, and learn about them and then see if they work there. So that's really hard. definitely do think that. I guess make sure, yeah, wasting time, spending all this time, like testing out different tools if you aren't going to be able to get it. Cause yeah, that, that, that will just suck.
Christina Carter (21:14) Yeah. Yeah. I think that's a good point, especially at the end, like you, before you do go look at them, make a short list of ones that actually are going to meet your like security data residency criteria. I guess I was just shocked at like the percentage of people who cannot use any sort of AI in their sales process.
And I, I do really worry about the teams who are using tools that don't really use AI very efficiently or at all within like their end to end systems.
Jasper Cooper (AutoRFP.ai) (21:41) An interesting one is the different model approaches from different vendors. Some will go with a very specific vendor and use that model only, right? So like some companies might do like Microsoft Azure and just do all of the AI there. That might be a consideration. But there's also the reverse consideration, which is vendors who can plug into all of the major AIs, right? The Claw, the Gemini and the GPT. And then they have the ability to switch to whatever's best for the use case. Cause as we all know, one second Claude's the best thing in the world. Next thing GPT is the worst and like, you know, it it rotates every day. So I think finding a vendor who can provide multiple models, same level of security could be like a really big benefit for people in a long-term set up because those same companies that don't like to bring on AI, it's going to bring on like a lot of procurement and you're probably going to end up with a three year contracts are really just making sure that vendor next year isn't still on an old model with potentially someone who's starting to lose the race.
Christina Carter (22:41) I think the end-to-end proposal management software category is going to change and probably absorb quite a few, I guess, other categories. I think that's where a lot of people are trying to move to. where do you, like, do you disagree with that? Like, where do you think we're going in this space? Like, what are people trying to go towards?
Jasper Cooper (AutoRFP.ai) (22:49) Yeah, I think we will see a lot more self-service where systems were historically like very hard to use as well. And not necessarily like it really required like an experienced person with like tens of hours of training in a particular software. I think we'll see like a lot more modern user interfaces, a lot more AI in the user interface so that it can assume things on behalf of users and really like bring down the
barrier to entry a lot, which means that I think a lot of proposal software will go more self-service. So the bid team providing it as an application, there's a lot of like frontline sales functions, investor relations, et cetera, that right now are on their own unless the bid is like above a certain threshold, right? Like if it's not worth 200K a year, you can't talk to us. Here's our library, like good luck with it. I think we'll see.
Bid teams, proposal teams provide a lot more value in the organization with only incremental effort by being able to give the keys to those different people in certain use cases, protecting them and knowing the level of accuracy. So allowing them to get those tools. So I think we'll see tools go from, yeah, we need it for 25 users to, yeah, we still have those 25 core users, but we're opening this up to a further 100 people that have no experience with.
this type of software. I think that'll change it fundamentally. Because of that, I think you'll also see a lot of like agent self-service, like being able to work with a chat bot that can answer technical questions, that can answer commercial questions. Like you're maintaining the super valuable content already to answer the most important questions that your company gets for like new customers, renewing customers.
spending all of that time, like your whole organization really needs that. So I think we'll see teams, you know, on the, we're talking about like 2027, even, I think we'll actually see a lot of teams move from RFP and proposal teams, like knowledge teams, information teams that will contain that. They'll still have that, that volume will still be exploding, but they'll now be expected to do this as well, just because of how well it will like naturally work in the organization.
and that 99 % of organizations don't have anyone formally in information management. But I think they'll start to learn, this is the proposal, this is working very well. I like this agent. I don't know why I don't like it. I like it more than that other agent I bought that is just sucking everything up. So I think that those will be some huge changes.
Another wave, think, will be like NCPs and things like that, right? So agents will more and more become consumers of the content that you have in the platform. I think you'll work through agents more inside of the platform. I think that'll be interesting. Like by then we'll look to have NCPs so you can use it through Claude and Gemini and et cetera, but also have agents within the platform. So I think you'll see that more and more and like highly specialized agents that just work on our function that focus on certain things require less prompting and can work on long horizon tasks. So that's something that like the model providers have struggled with and continue to struggle with is like this concept of long horizon tasks. Because if I get an RFP and it has 500 requirements just to call the search tools to find the content I need for that, that's going to be maybe 1.5 thousand different searches optimistically. And right now, the agents that are commonly put out there just aren't built for that. So I think we'll see a lot of very specialized agents for our use case.
Yeah, there's so much there, I think those are some of them. It's obviously going to be a huge focus on automation and huge focus on the knowledge base automation as well, particularly reconciling facts across different sources and keeping that up to date automatically. But yeah, I think, I think there's a lot of automation, but as we're starting to see as well, I don't know if you also saw this report, but about win rates going down from LUPIO with the win rates going down with the volumes going up as well as the time spent per RFE going up. And I think this is basically Jevons paradox at play of people are using AI. There's some people who aren't using AI. They're spending more time now. We're using AI. So we're bidding on more opportunities, but everyone else is also using AI. So the win rates are collapsing. So I think that's going to be like the most interesting stuff in 2027 is this going to be a real like race of like how much automation you can bring in and how to get an edge. we'll see a lot of, hopefully we'll see a lot of very interesting functionality of how you actually get an edge to win and less so automation, automation, automation, which is just the foundation I think of.
Christina Carter (27:52) No, okay. mean, so the knowledge, the knowledge management piece, I agree with you. think that especially if you are a proposal person listening to this, you're like, okay, where does my job go now? Like, what do I want from this? I think that's a really smart way of pushing it. If you enjoy that part of proposals, I think that's a great, great way of just being a moat in your organization. But also, I think that it ties back to that last piece of finding that little way, that little lever of winning.
I think the automation piece though is really important because right now we are seeing with, cloud skills and cloud plugins and it's so easy now to create a bunch of automations, outside of like regular proposal management software and then turn it into something that can just kind of be automated and you can decide where the gates are within that. And I think, I think that's what people are going to come to expect from their proposal software. I think proposal software that is just like, I'm just going to respond to this RFP or just respond to this questionnaire. It's just not going to be enough. I don't, I don't think people are going be worth or wanting to purchase a tool like that, even if it does have some more like security gates to it.
Jasper Cooper (AutoRFP.ai) (29:09) 100%. I think every vendor, like this is like the hardest I'll ever need to work. Like there's so many different players and everything like that. There's so much capability in the models. There's so much ability to go above and beyond like what's being delivered by generic tools, but you will need to get to at least that level and then faster pass it. So I think it's going to be like a really big year of hard work. And I think there'll be actually many of like
the tools that exist in the market just won't be able to make it through this precipice. It's going to be very, very important that you can bring those same workflows that you can from Claude into here or work with it in a way that makes sense. So yeah, I think the bar is absolutely going up.
Christina Carter (29:48) Yeah, yeah, which is hard for you. But great for me. Because that means my life gets a lot easier if all these tools work really well. But yeah, it's just last question. If you could sit down with a head of proposals or head of sales, like whoever's managing a team of people responding to RRPs, what advice would you give them when they are starting to evaluate tools?
Christina Carter (29:48) Obviously, there's gonna be some bias there, but still, what would you tell them for them to look out for that they probably aren't asking you right now, that you think that they should start asking you?
Jasper Cooper (AutoRFP.ai) (30:20) Yeah, I think an interesting thing happened in proof of concepts where all of us vendors kind of more formalized around like proof of concept. You come with like three previous responses, you upload them into the system and then you see the first draft, right? So it became less of like, let's evaluate the platform and what it does and how it learns like all of that kind of nuance. And it kind of just folded into first draft, right? Like what does the first draft look like?
I think that's where the evaluation can go. Definitely, you can do that. That's great. But I would definitely dive a lot deeper. And I would maybe try, like if you have the time and the resources to invest in putting a lot more content in there, a lot more gnarly stuff, trying to figure out what they use, whether that's tags or folders or hierarchy or whatever that looks like, use all of that. Plug in maybe a few different systems if you can do that during the proof of concept test the actual workflow, you know, God forbid, like actually click some buttons in the system and try and try like different features that sit off the maybe core beaten path, but you wouldn't use and leverage every day.
And just try and dig into it a little more. I would say that just kind of skim, do that, and then potentially be disappointed that when you went to use it in real life and a much more kind of, ⁓ built out scenario that that button wasn't there. it's not there or it drops in this way. And then the final part I would say is like the learning process is so important. Like there's some, there's some great quote around this that I can't recall, but around, it's not really about like where you start, right? It's like your incremental improvement. If you get 1 % better a day, then that just compounds so fast. And I think even with a quick test of a software, you can just see how much it actually compounds. And it's really simple. Like you upload one response, see how it generates, see that draft, work on it together as a team and prove it approve, approve, complete whatever their workflow, their loop is, and then run another one, maybe similar to it, where you expect learnings to come through, and then see how it learns and see its rate of improvement over time. And hopefully, you could expect that as a continuous rate of improvement as you got the product established. So I think that's a really important one, is that learning piece and one that systems are at risk of.
Some things there like systems will on average increase in hallucination rate, the more content they have because it gets like awfully confusing or if they have very little content, some systems will hallucinate more because they don't have so much to grasp at. So yeah, there's a lot of interesting considerations there when we're just talking about AI drafting. I feel like that's where a lot of the time is focused, but I think yeah, teams could definitely do better long-term by taking that as a component within the broader piece. And as we've spoken about.
Other pieces are now being automated. take a lot of time. Like they're important aspects of the process as well.
Christina Carter (33:06) Yeah, I'm, I'm with you. I think seeing a demo versus like having it actually look at, you know, your 3000 Q and A pairs that you've uploaded into it from like your, your past system are very, different experiences. And you're to want to like, see that for yourself. Like, I don't think any amount of research on, you know, the tools you want to use, like when you're doing your research and perplexity or whatever, it's not going be the same as like actually like trying it out.
Jasper Cooper (AutoRFP.ai) (33:35) Yeah, exactly. And it's an interesting thing for vendors because when that becomes like the proof of concept gate, then it's like, do I spend my time engineering on like actual customers with like, 30,000 content items or whatever, and all of this setup, or do you spend your engineering time just trying to win the proof of concept and make, first drafts from very little content look very good. And I think that's kind of some of what, what, was at play on the vendor side. So yeah, these are why you think about these kind of things. can have a system that's definitely over optimized for one thing. So make sure to just spread out and check everything that you need to. You might not need everything. You might just want one thing. Right. Just make sure to focus on that.
Christina Carter (34:13) Yeah. I say this a lot, but focus on the thing that you are having a hard time with. Like, what are you struggling with? Is it going to fix that thing that you're struggling with? Like, whatever that is. And if it doesn't, it does really something else that's really cool and you love it, but it doesn't actually fix whatever that problem is that you're having. it's not, it's not the right software for you.
Jasper Cooper (AutoRFP.ai) (34:26) Exactly.
Exactly, and your evaluation criteria is your evaluation criteria, right? Don't, don't let us vendors go, no, you need to do this. And then you need to do this. then like, do that. Great. Follow some instructions. Very useful to see the platform as they want you to see it. But then maybe step a little bit outside the guard rails and make sure that your requirements are actually being met. Because of course they're going to instruct you to whatever benefits them for the fastest close. Right. Um, so yeah, all good things to be, to be mindful of.
Christina Carter (35:04) Yeah, it's good for both people. It's good for the vendors and for you because you're not going to churn in a year. You're not going to waste your time. You're not going to waste your money. It's the right thing to do. All right. I think that's the end. Those are the end of my questions. Everybody go check out the category report and just go see Auto RFP in there and see how they're doing. And yeah, I will talk to you all soon.
Jasper Cooper (AutoRFP.ai) (35:10) Exactly. Exciting times.