Recording:
Transcript
**
00:10**
Daniel Englebretson
This meeting is being recorded. Good morning.
**
00:31**
Simon Walmsley
No worries. Just. I must admit I’m not. It’s been a while since I’ve used Zoom normally on Ms. Teams, but. Yeah, just. I’m joining through the web browser, but just getting myself familiarized. Yeah, no worries. What time is it for? For you, 10:00am all right. Okay. Yeah, 3:00pm for us over here in the UK. Are you in. We’re based in Stockport, which is just outside of Manchester, so in the northwest.
**
01:16**
Daniel Englebretson
Nice.
**
01:17**
Simon Walmsley
Yeah. And yourself?
**
01:20**
Daniel Englebretson
Where.
**
01:20**
Simon Walmsley
Where are you based?
**
01:22**
Daniel Englebretson
In North Carolina, in the center of the state. So almost smack in the center of the eastern coast of the US though.
**
01:29**
Simon Walmsley
All right. Okay, cool.
**
01:41**
Bhumika
Hello.
**
01:43**
Daniel Englebretson
Good morning.
**
01:44**
Bhumika
Good morning.
**
01:46**
Simon Walmsley
Hi, Bhumika.
**
01:47**
Bhumika
Hello. Sorry, my computer is running a little bit slow, so I’m late. I had to restart.
**
01:55**
Daniel Englebretson
No worries.
**
01:57**
Bhumika
Is Steve join us then?
**
02:02**
Simon Walmsley
I haven’t seen Steve today, but then that’s not unusual because he’s in a different part of the building to me. He hasn’t told me otherwise, so I’m expecting him.
**
02:13**
Bhumika
Let me just. Messages. Otherwise we can get started.
**
02:18**
Daniel Englebretson
Awesome.
**
02:21**
Bhumika
Did you guys go through the introductions?
**
02:24**
Daniel Englebretson
Not really. We just kind of said hello and here we are. Yeah, happy to do some intros though.
**
02:32**
Simon Walmsley
Yeah. So I work for Hughes Safety Showers and I’m the proposals engineer. So I’m the person that’s sort of hands on dealing with these RFQs. Yeah. And Steve from our side, I think he’s been involved in some of the AI Council work with. Just.
**
02:54**
Daniel Englebretson
Right, okay. Okay. Well, Simon, you’re going to be a phenomenal person for this conversation, so we can wait, but Mika, or we can get started, whichever you prefer, can get started.
**
03:07**
Bhumika
He just messaged me that his current meeting is going over so we can get started.
**
03:12**
Daniel Englebretson
Okay. I’m putting a link in the chat for a lucid board. You’re welcome to follow along inside. Lucid. Or if you just want to watch my screen, that’s. That’s totally fine too. So I’m going to share my screen as well. And let me, let me orient us a little bit with what we’re getting into so that you know kind of what we’re looking at. So before we get started, first thing I want to call out is everything up here at the top doesn’t matter for this conversation, so don’t worry about what’s at the top. We’re mostly looking down here in the middle. And so just to orient us really fast over here is a first pass at quantifying some of the things we’re Trying to improve and reduce. So I’ll come back to that a little bit.
**
03:58**
Daniel Englebretson
But the main focus of today is the workflow itself. And so what I did is I took what was supplied to me and broke it. And here’s. These are basically the names of the files that were given to me. And I basically broke them out and described the files. And then from those files I tried to describe what’s the process. And so the black right here is the central kind of process that we’re going through. But the main steps of that process right here are broken further into this green right here. And so I want to highlight, especially for you, Simon, all I was given were the documents and nothing about like what the documents are and how they’re related. So these are guesses. I think they’re intelligent guesses, but they’re guesses. So.
**
04:45**
Daniel Englebretson
So one of the main goals for today as we go through this is to make sure that I actually do have it right, that this is the flow that it would go through. And, and then once we have confirmed that this is actually the flow, I want to talk to you about where I call it the friction. Where’s the friction in the process? What drags you the most, like what sucks your time the most when you’re doing this? And then also where’s the opportunities for AI to basically streamline your efforts a little bit? So we’ll talk about that a little bit more.
**
05:21**
Daniel Englebretson
But it’s all centered around kind of the first statement I need to make here is, I think the problem we’re aiming for is you receive a project inquiry or a materials requisition from a client, and then it has a whole bunch of technical requirements. And then you have to reference standards and data sheets when you do that. And your objective is to demonstrate that your proposal is a good fit for those requirements. And then ultimately you want to close the deal. So that’s what I think we’re doing. So let me pause since I know we just had Stephen jump in and say hello and see if we want to orient Stephen real quick. Good morning.
**
06:04**
Simon Walmsley
Okay, Steve.
**
06:09**
Bhumika
We cannot hear you if you’re trying to talk.
**
06:15**
Daniel Englebretson
Guessing we’re getting audio set up here.
**
06:20**
Simon Walmsley
But I just want to jump in though, Daniel, and say I have had a review of the documents that you provided, the workflow. And you know, if you’re saying that you didn’t have any context other than the documents you received, you’ve done a very good job interpreting the workflow pretty much there.
**
06:40**
Daniel Englebretson
Good, good. That’s, that’s good news. I mean, it’s somewhat obvious as long as you kind of know what’s going on, but I can see that it’s complicated. So, so given that, then the, when we think about designing a workflow. So taking a step back, we’re about to have a workshop together as a team. And my role in this workshop is twofold. It’s one, to teach, to convey concepts around how can AI help you do your job better? And then to basically teach you how to do that for yourself. And so, but for all of this to make sense, we needed a real thing to look at, so we could be looking at a real flow, talking about these things. So the real flow that we picked is this flow for a lot of reasons that Bameka basically decided. And so just right.
**
07:33**
Daniel Englebretson
Brought to me, we’re going to do this flow. And so, and so when we think about anything that we’re doing alongside AI, it starts with, okay, what am I reacting to? What’s the signal? And then it’s like, okay, well, what’s my objective that I want to achieve? And then like, basically, what are the steps that I have to take to achieve that objective? And then what’s, what does a good outcome look like? And so it doesn’t matter what we’re doing. It all basically follows that flow. And so once we break down, okay, signal, objective, steps, outcome, then we can start to say, all right, well, in what way can I help us do this? And so, and so there’s two kind of sides to this coin. One of them is reducing friction in the process. So, so, like, where is data getting stuck?
**
08:19**
Daniel Englebretson
Where are you having to spend a bunch of time? Where is there some decision that’s being made in your head that’s not really written down anywhere? You know, what’s basically just slowing everything down. And are there pieces of that we could mitigate with AI collaboration? And then secondly, how can you actually make the result of your work better? And so that I use this acronym, streamline, enrich, accelerate. So can we reduce the cognitive lift? Can we make it easier for you to do it? Can we make the outcomes better than what it would have been otherwise? And can we accelerate the pace of this whole thing? So it’s important, when you’re talking about accelerating, a lot of people just think like, oh, I’ll just get it to write my proposal a little bit faster, or something like that. It’s more about the entire Flow.
**
09:04**
Daniel Englebretson
How can we speed up the pace of this entire chain? Not just a piece of the chain, because at the end of the day, if you just, if all you did was sped up the cross reference, yes, that’s valuable, but it’s still not, it didn’t really speed up the whole thing. It’s just that one little chunk. And so for us to be able to get our arms around that, we have to take a step back and say, okay, well, what connects to what connects to what to look for. Where can AI help us do this? So that’s basically what we’re doing. And at the end of it, we have been successful and we can show the way you were doing it versus the way you are doing it now. After all, this is better, faster, cheaper, whatever, Right, Whatever that is.
**
09:46**
Daniel Englebretson
But I don’t want to tell you the goal is to make it cheaper. I want to tell you that we’re going to measure it with streamline, enrich, accelerate, and we’re going to figure out what the goal is. Right. And so, and I think it’s important to call this out because a lot of times this takes the tone of like cost reduction, which certainly is part of it, but there’s also value delivery. How can we make the proposal better? Like, how can we, how can we look across it and actually make the response better? And some of that is in pace because if you get an RFI and you can respond like now, you’re going to beat the competitors in the response.
**
10:19**
Daniel Englebretson
So some of it is pace, but that’s, I’m really trying to push us of like from the customer’s perspective, what’s actually valuable to them, not just in terms of reducing the time it takes to do a thing or whatever. So it’s both sides of it, right, that we’re thinking about. And so what I’m hoping through our engagement is that these things like streamline, rich, accelerating, friction start to make more sense because we’re looking at them through the lens of a process. And so for that to work, we have to understand the process. And so when we get to the workshop, what I owe you is a AI assisted workflow that does this thing. And then we’re going to build them together so you could kind of follow what I’m doing.
**
11:02**
Daniel Englebretson
And then when we’re done, you’re going to have a flow that takes all these inputs and give us, gives you an output. And so, and so the idea is over doing this a Few times. Eventually you’ll get the mechanics of it and you can just do it. Maybe that’s the first time, maybe that’s the 10th time. That’s what we’re trying to figure out. That being said, if all of the inputs that come into this, that signal, it’s time to get started. Are these documents basically that we laid out, or something like this, where you get these specs and you have a representative quote and you have to do the cross referencing and all these different documents, those are the potential inputs that are coming in when you receive the signal. Then we just have to start saying, okay, what’s happening?
**
11:46**
Daniel Englebretson
You’re looking at some stuff, you’re asking some questions, you’re getting some answers and you’re doing something with that. So we might find that some piece of this AI adds no value. We might find that. And so it may not be every single piece of this, but where could it, and so as we think about that, it’s about where could it speed you up? Where could it give you a better outcome? Where could, where could it make it easier for you to just execute?
**
12:14**
Stephen
So.
**
12:17**
Daniel Englebretson
Part of that is in what documentation do you already have around your processes? So like, maybe you have templates for what your proposals look like, or maybe you have quality checks that are written down for like every time we do a cross reference, we look for this, this and this. You know, maybe there’s stuff of that already written down and we can use that. Or maybe some of that stuff doesn’t exist and we need to create that to make this happen. And so I’m going to pause in just a second, but the last idea I want to land on you before we start thinking about this is it basically all comes down to what questions are you asking of what data and what are you doing with the answer.
**
12:55**
Daniel Englebretson
And so you, the expert, you know, okay, I need to know A, B, C, D. And you know, I need to look at these things in order to answer those questions. And then, you know, once I know abcde, I can go do this thing. And so we just, we’re basically trying to boil it down to what questions are getting asked of what data. And then we got to go figure out, do we have the data? Can we even ask the question what will, you know, we have to figure all out later. So let me pause there and see, you know, in terms of just kind of grounding and what we’re up to for the, for the flow and any questions about kind of the general Scope of what we’re. What we’re trying to look at here.
**
13:38**
Simon Walmsley
No particular questions from my side at the moment, but I do agree with the point you’re making that not just the speed that this could improve upon, but putting together a more thorough package at quote stage to try and eliminate some of the time that we invariably spend when the customer comes back with questions. There’s always after the quote is submitted, we go through. Customers have different acronyms for the phase, but it’s like a technical alignment phase. So if we’re able to extract a lot of the data out of all the various documents we receive at quote stage and we can put together a really comprehensive compliance matrix, it’s kind of demonstrating that we’re one step ahead. Already we’ve considered almost everything that you’ve thrown at us, and this is what’s relevant and this isn’t relevant.
**
14:36**
Simon Walmsley
And I think that would certainly help set us apart from our competition in demonstrating know superior knowledge and competence.
**
14:47**
Daniel Englebretson
Okay, okay.
**
14:49**
Stephen
And just to add to that. Sorry, the providing that value, I think, was a good way to describe it. Daniel. I like that. We’re not actually chasing a cost cutting measure. I think that’s just something internally we had to fight with. But I think the value bit is what it’s at. And to kind of add to Simon’s point. And I’m really happy you picked up on all this. Providing that very professional response with a lot of the questions preempted with our, you know, excellent technical understanding, but at speed will probably initially surprise the customer because they won’t be used to it, I don’t think. And it’ll certainly give us a. Sorry, you might correct me on there, but it’ll certainly give a customer like, oh, these guys do know what they’re talking about and how quickly they come back, because it could.
**
15:41**
Stephen
These things can take weeks and weeks, can’t they, Simon? And just to get this, you know. Yeah, yeah, there’s a lot of value in there. So thank you.
**
15:50**
Daniel Englebretson
Yes, yes, we’re definitely on the same page. And so. And so as we break down, the goal of this meeting today was one, make sure that I’m generally correct. And two, you’ve already given me two, which I kind of sticky noted on here. We’re looking for. All right, as you. As you’re doing what you’re doing, you know, you’re reviewing product project specs and data sheets and you’re analyzing requirements, you’re doing these things, what would be really nice? What would make this thing way better for the customer. And what would make your life easier or better if the machine could take some of that off of you? Basically. So it’s like, okay, so I know it takes me 10 hours to read all this stuff. And the reason why is because it’s really hard to read the fine print or whatever, right?
**
16:35**
Daniel Englebretson
So whatever that thing might be. And so as we spot those things we’re looking for, like, basically if you were dreaming, essentially, if you had a very capable intern who could do all this little stuff for you and just bring it back to you, what would those little things be? You know, and that we’re just trying to find those and then why would that be cool not only for you, but also for the customer? And so what you just said about, you know, higher quality responses, demonstrating expertise and then surprising them because you got back so fast and conveying, oh, wow, these guys know what they do. That’s exactly the kind of insight that we’re looking for in this. Because that mentality will shape how the system is put together so that we can make sure that we’re trying to achieve that.
**
17:18**
Daniel Englebretson
You know, so when we, when I take the first pass after this conversation today, I’m going to try to basically scaffold one out that demonstrates these things so that people can build on themselves so that you can scale, see it happening. And then over time, yes, you could just leave it like that. But the real value comes as your operators. As you guys are doing your job in the future, you’re like, oh, what if we did this too? Or what if we change this? Or what if we made this change? You know, so the idea is that like, yes, we can build a system that does it today, but we want to build it in a way that like, as you have more ideas of how to make this thing better, you can actually make it better. So that’s where we’re trying to go.
**
17:56**
Daniel Englebretson
So, so the main thing that I wanted to achieve coming out of here, other than grounding ourselves, is we want to come up with a few questions. And maybe this is the whole team, but we wanted to come up with a few questions of what do we need to ask anybody else involved in this process before we have the workshop of what else would be nice to reduce or what else creates problems or what else creates value? Because what I would like to do is have like two or three questions that we could send out to the team and maybe you are on the team and we can just answer it that says okay, what. What really slows us down? What would the customer really find most valuable?
**
18:32**
Daniel Englebretson
And whatever that other question might be that we’re going to come up with here so we can ask a few more people involved, even if they are or not in the workshop, so that when we build the first pass at this thing, we get the best possible outcome from the first pass. You know, that’s. That’s basically what we want to do today. And so we can. We can go down the list and just kind of say what comes to mind. Step one, step two, step three. Or you might already have some ideas that you just want me to document. And then what I’m going to do is pull all that feedback together and make that our first pass. So that’s what we’re trying to do with the 40 minutes or so we have left.
**
19:06**
Daniel Englebretson
And I’m happy to kind of take it step by step. Or if anything’s jumping out at you that you want to hit right out the gate, we can do that, too.
**
19:15**
Bhumika
We should also think in mind, like, look at from future perspective, as I think Daniel, you and I talked about, and Steve and Simon, I emailed you about that. We should also take, like, while we are building it. I don’t know which complexity that is. I’m assuming you are getting, like, low, medium, high complexity. Like, some of the documents might be, like, very high. So maybe we should start with a medium and then, like, take a. Take a high one. Again, I don’t know the details of it, which, Simon, you probably can provide, like a complex one and then try to.
**
19:54**
Bhumika
Once we have, like, the steps together of what we need to ask and what we need to pull up, we should also, like, take some of the previous ones that you have done and try to run it through here and compare the results too.
**
20:09**
Stephen
I think one of the challenges. Go ahead, Simon, you go first.
**
20:14**
Simon Walmsley
No, I was just starting an agreement.
**
20:16**
Stephen
Oh, yeah. I’m just going to kind of. I don’t do this job. Simon does this job. But as I was messing around with my very amateur prompt, I was starting to learn the challenges, what Simon has from an outsider looking in. I don’t know if there’s too much detail at this moment in time, but what we do have, if we get the highly technical ones, we get a suite of documents. And inside of those documents, they refer to other documents who refer to other documents. And it’s kind of following that breadcrumb trail of these. These references that I struggled with. It was kind of, okay, refer to document, blah, blah. For this particular bit of detail.
**
20:55**
Stephen
So I just wondered, one of the outcomes would like to see is that being able to understand what that from a kind of almost like an audit perspective, what that trail looks like and make sure that it’s following through. I think that was one of the issues that triggered this conversation. The first place is we missed a document deep within this trail. And it wasn’t a document we missed, it was one particular spec within said document which kind of like five documents deep. I’m not sure if I’m making much sense now, but that was kind of a challenge at this top bit for me.
**
21:32**
Daniel Englebretson
Okay, that’s a great example. That’s a perfect example. And so just to keep us grounded in this, one reason why it’s a perfect example is because the cognitive lift to go, go, go is heavy. Right. Also the time it takes to go, go as long you can compress that. But also enriched outcome means oh, you found that one detail that you might have missed otherwise that because you found that response is higher quality. And so that’s when they hit all three. Right? And that’s perfect. And so that’s exactly the kind of insight that only you guys are going to have. So, so I try to capture that as. Okay, so what makes something more complex? Is there such a thing as small, medium large documents that refer to documents drives complexity. So we need a feature that is generating, they’re called maps of content.
**
22:15**
Daniel Englebretson
Like this is related to this, related this when it comes in. And what would make that nice is looking for one particular spec within a document or whatever. We’re just trying to generalize, but that’s a perfect use case. So step one, review project specs and data sheet. You already kind of called that out. Is there anything else that would drive a heavier complexity or slower turn time or it requires somebody who knows who’s like very experienced to look at it versus less experienced or anything like that’s happening here at the beginning.
**
22:52**
Simon Walmsley
No, it tends to be just the amount of and detail of the documents that we receive. And you know, sometimes if we’re dealing with a distributor, we very basic data sheet and that’s it. And sometimes the data sheets will refer to other document numbers and we may have to go back and ask for those. Sometimes the distributor might not even have them available and they’ll say I’m sorry, I don’t have this available. So you have to make certain assumptions and or rather maybe you know, identify that as a deviation in itself. That such and such a document was not available. And then the flip side is if we’re dealing with an EPC contractor and a lot of these guys are the ones that have actually created these documents, you know, you can have maybe 100 specification documents that you get sent.
**
23:47**
Simon Walmsley
And you know that’s going to be a labor intensive, you know, process for doing a quote. And so that’s how you kind of get a feel as well. You know, is this going to be a quick and easy quote or is this going to be something that’s going to take lot of time?
**
24:04**
Daniel Englebretson
Okay, you said when getting a certain type, what was it? Fec, what did you say?
**
24:09**
Simon Walmsley
You had an acronym rfq, request quotation.
**
24:14**
Daniel Englebretson
But it was a certain type like that where there were.
**
24:18**
Simon Walmsley
Oh, epc. Yeah. So that stands for engineering, procurement, construction. So that’s the type of large contractor that does the complete turnkey solution. So they work directly for the end user, such as Saudi Aramco or adnoc. And they’re the guys that will design the new facility. They’ll do all of the engineering, they’ll procure everything, and they will have engineers on the ground building the plant right up to commissioning.
**
24:51**
Daniel Englebretson
Okay, okay. So, okay, so I’m following so far. So you get all this stuff in, you got to go through it all at the point that you are kind of intaking materials, finding what you’re missing, going through it all. And then you’re now officially analyzing. You’re kind of coming to the end of the analyzing of the requirements. What are some of the, like, if we missed this, it would be a huge mess or like the big risks of like, you know, basically the reason why we need humans looking at this and thinking about this is because if we miss this, we have this problem. Like what are some of those kind of red flags or gotchas or pitfalls that we would need to be thinking about?
**
25:37**
Simon Walmsley
So one of the big ones would be compliance with Atex standards A, T, E, X. So that’s a specification related to suitability of equipment to be used in an explosive atmosphere. So although it’s a safety shower, we do end up fitting various electrical items. It could be alarms, junction boxes, switches, these kind of things. And the specification of those items needs to meet the Atex requirements for that particular project. Because if you get that wrong, then potentially you’re very serious. Yeah, you’re supplying equipment that could ignite and cause an explosion one of these plants. So that’s probably One of the biggest. You’re then looking for compliance to material grades. So certain types of stainless steel, the grade that it needs to because that’s to do with corrosivity, there could be various chemicals or gases that could react with the materials that were providing.
**
26:52**
Daniel Englebretson
Oh, so I understand this. If, is it a case that you sell someone something that’s not compliant, it’s your job to know that it’s compliant and then now there’s a safety issue, or is it a case that you try to sell them something that’s not compliant, they realize that and so you lose the bid.
**
27:11**
Simon Walmsley
Yeah, a bit of both. And I mean, to be fair, the. When we get it, once we’ve submitted our proposal, I mentioned previously about a sort of technical bid analysis, a phase that we would go through with the customer. And that’s where you tend to get a whole host of questions and some of them will, you know, very basically say, do you comply with Atex requirement? Blah, blah. And you’re given that opportunity to say, yes, we confirm or no, we don’t. But at the very start of the bid, you need to make sure you’re not wasting time selecting equipment that’s not suitable. And these sort of technical reviews that happen after the bid submitted, they can be quite exhaustive and for the reasons I just mentioned, you can see why they need to be.
**
28:07**
Simon Walmsley
But again, demonstrating compliance and understanding of that at the bid stage is certainly going to give our customer some reassurance that we know what we’re doing and what we’re offering is suitable.
**
28:23**
Stephen
I think just for context as well, Daniel, when we kind of bidding for the jobs we’ve been, we’re providing a safety shower, but we are being treated the same as we would be if were providing a, I don’t know, an oil rig, full oil. We get treated exactly the same. It’s not, you know, there’s no dumbing down of the requirements. Probably another one there. Simon, these paint specifications.
**
28:48**
Simon Walmsley
Yep. Yeah.
**
28:52**
Daniel Englebretson
Okay.
**
28:54**
Simon Walmsley
And yeah, that kind of explains, Daniel, why we can end up with some RFQs that have over 100 specification documents, because that same suite of documents has been sent out to a supplier who’s providing a really complex bit of machinery or equipment, some sort of turbine or compressor or. Whereas on the surface, the safety shower may not have that exact level of complexity, but they still want you to say, yep, we comply with everything, you know, even if it’s certain things are not applicable. We’ll get that same sort of level of detail.
**
29:37**
Daniel Englebretson
Okay, okay, that’s super helpful. So then, so then when you are, you now feel grounded in the whole thing, you’ve got a grasp on the thing and you’re trying to actually engineer the proposal, you’re putting that together and you’re kind of configuring that proposal, if you will. Is it like a modular menu of stuff? Is it totally from scratch every time or what’s happening when you feel like you’re ground? Or is iterative, like there’s a base layer of kind of putting it together and then you’re kind of making changes? Or what’s the process of actually. Okay, now I know what I need to put into this thing. So I need to start putting the proposal together. How does that, how does that actually happen?
**
30:25**
Simon Walmsley
Yeah, so I think that comes down to the sales engineer’s knowledge of our product range. You would quickly identify the type of safety shower that would be suitable. And when it comes to configuring our offering, we use Salesforce CPQ to build out our quotations. It’s not particularly intuitive in that you’ve selected this type. Say like we have a tank shower or we have a tubular shower. If you’ve selected the top level family of products, we don’t then necessarily have a sort of user friendly selection tick box type exercise of. Right. Well, we could add this fitting or this fitting. It all tends to be reliant on the individual’s product knowledge of what things go together.
**
31:30**
Daniel Englebretson
Okay.
**
31:33**
Simon Walmsley
Between us as a department, we got a lot of knowledge and experience and we’ve also got our engineering departments to draw upon. Because if there’s specifications that, you know, an individual might not be confident in saying that we comply with, we’ve got, you know, the other departments that we can draw on their knowledge.
**
31:57**
Daniel Englebretson
Okay. Okay. So I’m glad you called that out. So one way that a system like this can help is as we identify the expert knowledge that people have that junior people don’t have. It’s about making that export knowledge accessible to a junior person in the form of an expert knows you need to ask this question when you’re looking at this data and the junior person doesn’t know that. And so when the expert is going like pouring through the catalog, you got to go here. You got to hear what they’re really doing is like, I know from experience. First I gotta be like, which family?
**
32:34**
Daniel Englebretson
And then I gotta know which category and then I gotta know, you know, so if we can get that type of thing written down, even if the system can’t automate that, it can automate telling the person, don’t forget, do this, this, this, you know, and so that’s part of the streamline enrich, because that’s a case where your organization, someone knows how to do the thing, but someone might be having to reinvent that wheel every time because they don’t know how to do that thing.
**
33:00**
Daniel Englebretson
And so as you’re thinking about that expert knowledge, is there any time where like particularly really one way to look at this is there’s no way a junior person without experience would be able to answer this question like what are those questions, you know, that really require somebody who really deeply knows the system to be able to do it?
**
33:26**
Simon Walmsley
Yeah, yeah, that’s a good point.
**
33:28**
Stephen
It’s, it’s a very niche product in itself. You know, Safety Shower, obviously it doesn’t really do the complexity any justice in its title. It’s very niche. Before you had Atex, before you had other compliance things, you know, we’ve got flow rates, you’ve got all sorts of different variables. However, we do have very good data sheets which distinguish some of these niche requirements. And I think we could, if we could upload that as a bank of knowledge to get us 80% there with a 20% check from assignment, you know, at the end, just to verify that where it’s taken us is fine. Having played a little bit that myself, it did chat GPT did a fairly good job of kind of comparing the spec and the day machine coming out with a, an option.
**
34:28**
Daniel Englebretson
So, yeah, so in that, so given that you play with chat PD sound, that’s going to make this helpful for sure. Because basically when you’re using the LLM to do this stuff, the difference between just going to chat GPT and what we’re trying to do is basically sequencing questions to ask the LLM so that the LLM can actually answer it. And another way to think about that is you’re trying to make the question less complex. So instead of being like what product should I quote? If you can break it into 10 less complex questions like what should the flow rate be and what should the product, what should that, you know, then the likelihood that it gets that answer right is much higher because the answer is a much simpler question.
**
35:11**
Daniel Englebretson
And so what we’re trying to get 80% of the way there might be. Okay, how can we reduce the Complexity of what we’re sifting from this by chunking it into different questions. And so to that end, like you already mentioned, okay, if we could just better mine the data sheets to ask these questions that get us 80% of the way there. It would, I’m guessing is coming from like, you know, we’ve mapped the content, we’ve understood the respects, now the specs, now we have to mine the sheets to look for these answers. So even things like, okay, could I even just filter them all down to the ones that only talk about flow rates of X or whatever that might be.
**
35:50**
Daniel Englebretson
And so those kinds of things are what we’re thinking about on the friction side also as we’re thinking about this here, on the customer value side, is there a case where someone could go find a data sheet? That’s correct, but it’s not the best one. You know, is there a case where it’s like if you had quoted this and not this, they both would have worked, but this one is a better one for this scenario. Is there a case like that or is it always going to be one answer?
**
36:22**
Simon Walmsley
I think there can sometimes be some nuances in, you know, you could have like I’m thinking, Steve, like a tubular shower would work, but if it’s for a hot environment, a self draining tubular shower would.
**
36:36**
Stephen
Yeah, yeah.
**
36:41**
Simon Walmsley
Yeah.
**
36:44**
Stephen
Yeah. I think what you’re saying that there could be multiple solutions to the single problem. Yes, for sure. And I guess it depends on what the customers, sometimes they come in with a, you know, they’re almost there in terms of what they want. Right. So they’ll say a tank shower. They’ll say, oh yeah, the nuances is probably that 20%.
**
37:09**
Daniel Englebretson
Okay, okay. All right, that’s super helpful. All right, so we’re preparing the technical slash commercial proposal. You know, sounds like that’s a CPQ process in Salesforce. And so is the final document that comes out basically. Is that, is that basically compiled from a CPQ process or does a proposal engineer aggregate a bunch of information, put it into a thing and then do more work on that proposal outside of the system?
**
37:47**
Simon Walmsley
So the example quote document that I think you were provided with a couple that comes straight from our CPQ system. But the way our CPQ system for Hughes is set up is we have effectively a list of, you know, could be two or three thousand products from finished safety shower products to individual components, you know, optional fittings that get added to the shower. But it’s down to the, you know, the sales engineer to know what it is, they want to build into the quote. And so it’s one search bar at the top of the screen and you start to type in the code of the product that you need and there it is. Select it, add another, you know, and repeat. So it does rely on knowledge of people to know what it is that they’re looking for.
**
38:45**
Simon Walmsley
Rather than, I’ve selected this type of product, for example, a tank shower. And then I’m presented with a whole host of options that are only specific to a tank shower that I could select.
**
38:56**
Daniel Englebretson
Okay.
**
38:57**
Simon Walmsley
Literally just a big list of products. So a person in theory could potentially select the wrong option for that particular type of shower.
**
39:08**
Daniel Englebretson
Okay. Okay, let me write that down. Okay, so in that case, because I think there’s some nuance in the roles here. You have a salesperson who knows what they want and then you have a proposal engineer doing something in this process. Or what’s the difference between the roles?
**
39:34**
Simon Walmsley
Turn a phrase, it’s me. So as a proposal engineer, I’m reviewing the technical specification that comes in, I’m configuring the product, I’m creating the quote, and I’m then following it down and hopefully closing the deal and securing the order. Okay, so that, so it isn’t a sort of you hand it over to this department or this person. It, it’s managed by, you know, by one person.
**
40:06**
Daniel Englebretson
Okay, so if this is a question you don’t want to answer, you don’t have to answer this. Are, is your incentives? Are, are you incentivized in some way connected to this process? Like is this the efficacy of your work drive incentives for you, or are there any incentives in this process?
**
40:26**
Simon Walmsley
So the overall, you know, incentive, you know, you’ve obviously got the company goals and you have individual performance as well, your own goals. And anybody who works in a sales type role is incentivized with our SIP program. So, you know, there’s a sales bonus structure and you could kind of, if you look at it in the granular detail of what we’re doing here, okay. That, you know, you might not think, well, what’s the incentive to get this correct? But ultimately the incentive is you put together a good proposal. It’s something that the customer is more inclined to go towards compared to, you know, a competitor who’s not so hot on this. And you know, if you know, you’re getting the award, then you’re in a better position to achieve your own targets and get your sip.
**
41:23**
Daniel Englebretson
So, okay, that makes Total sense. And so I think you probably can tell I’m asking the questions. And so for you, then, as the expert in here, how do you. How do you know you did a good job? Like, I don’t know, maybe you don’t want to say this out loud, but it’s Friday, you’re about to get off work. You got to get this thing done. You’re like, maybe I should have done a little bit better. Better. Or maybe it’s like, man, I nailed that one. Is there, is there a spectrum of, like, you know, I know I crushed this one versus I had 20 I had to do, and we just needed to get this one out. Is there a spectrum in there?
**
41:55**
Daniel Englebretson
And, and if there is, how do you decide which ones you’re gonna, like, give the extra love versus, like, you know, it’s Friday at five. You know, is that. Does that exist in any way?
**
42:11**
Simon Walmsley
That’s a hard one to quantify other than feeling of knowing that you feel on top of the RFQ and you’ve got a good understanding of what the requirements are. I often think that, you know, sometimes when you’ve put a quotation together and you’ve got, you know, half a dozen or a dozen deviations that are listed out in the quote, you might think to yourself one hand, oh, well, you know, look, I’m offering a quote to a customer here that’s, you know, I’ve got a dozen things saying where we don’t comply. But actually, I think that’s a good proposal because that’s demonstrating competency and knowledge of what their requirements are. And I think sometimes it’s just as important to highlight where you’re not complying as opposed to just saying, yeah, we comply with everything kind of thing.
**
43:06**
Simon Walmsley
It’s, it’s that demonstration of knowledge to the customer.
**
43:13**
Daniel Englebretson
Okay, okay.
**
43:14**
Stephen
That makes contact. Daniel, just on the. So indirectly, you know, Simon doing a good job on all of these, you know, it for incentives. It does affect everybody’s incentives because, you know, we’ve got, I think We’ve got a $24 million pipeline, and it covers 110 projects, which you can’t get anywhere near those projects. The level of detail are these documents taking. And we’re kind of looking at it from a perspective. If we get 1% more from that pipeline, it’s another 240k into revenue. And by having a successful outcome to this, we’ll give Simon the time to leverage his skill set rather than going through documentation, but leverage his skill Set, providing good technical support to the customer, engaging with customer. We’ll get more involved face to face with customer. We’re quite certain that’ll be a big benefit and will help us.
**
44:24**
Stephen
You know, I think more than 1%, we’re going to set a 1% target, but I think there’s more to be had there. So yeah, we’re all incentivized by this. I think it’s kind of what I’m saying because we think there’s a, there’s gold in them there hills.
**
44:38**
Daniel Englebretson
Yes, yes, that, that’s very valuable nuance. Thank you. That’s definitely going to shape how I think about this because as we start building features into the flow, you know, you’re making trade offs and that type of trade off, like they’re called heuristics, a rule of thumb is how you make a better decision based on expert knowledge. Right. Like we make these trade offs for these reasons and that’s part of what makes you an expert. So that’s really, that’s really helpful. Okay, so recognizing we have like 15 minutes or so left, let me just kind of sum up the next three and then we’ll go from there. So I observed that someone’s creating a spreadsheet that’s kind of cross referencing a bunch of stuff in order to check their work. Is that correct? That that’s what’s happening?
**
45:25**
Daniel Englebretson
Somebody’s basically logging a checking of their work or what’s happening in the compliance matrix. It was like a little Excel sheet that was given, I think.
**
45:39**
Simon Walmsley
Is that Excel sheet? Was that one of the.
**
45:44**
Stephen
That was me, yeah. That’s what I managed to generate.
**
45:49**
Simon Walmsley
Okay.
**
45:50**
Daniel Englebretson
Oh, okay, okay.
**
45:52**
Bhumika
So.
**
45:52**
Daniel Englebretson
Okay. All right, so this is. Okay, okay, that makes sense. So, so this was more of like an assistive thing for you to like check the work with GPT versus like hard coded into the.
**
46:05**
Stephen
Okay, this is amateur hour, Daniel. This is messing around with.
**
46:11**
Daniel Englebretson
We gotta learn. All right, that makes sense. Okay, so that, that could be a byproduct of the process overall. It’s not necessarily a step in the process today. Okay.
**
46:25**
Stephen
No, yeah, but, okay, but I think it’s something which would be useful for both us on customer to say, hey, this is kind of, this is a, this is where we’re at. And I think as another outcome of that then it was if we had gaps in compliance, then I was, I managed to generate questions for the customer which said, okay, we’ve got this and this, but what about this, and this.
**
46:53**
Daniel Englebretson
Okay. Okay, let’s have more.
**
46:56**
Stephen
And then that way to an email format as well, just so I can kind of just get that away.
**
47:02**
Daniel Englebretson
Okay. Okay. So. So because you have to be able to trust what you’re putting out to the customer, you’re going to want to feel good that whatever you put out is right.
**
47:13**
Stephen
Yeah.
**
47:14**
Daniel Englebretson
Today you’re doing all the work yourself, so you feel good about it. But when a machine steps in, pieces of it, the machine’s doing. So now you got to feel good about what the machine’s doing. And so if you need to feel good about what the machine’s doing, where we need to understand what would you need to see to know the machine did good work. And so it’s like, because you’re not going to just cut and paste, at least not anytime soon, because you want to build confidence the thing is working. And so it’s like when it comes in and creates the map of content, says, here’s all the files and here’s related, and here’s all the specs that were asked for.
**
47:46**
Daniel Englebretson
And like, here’s the line in the other file where the spec was found, and here’s the sentence that it said, you know, being able to kind of service all that back. So that when you look at it, Instead of reading 100 documents, you read a cover sheet that says, here’s all the specs, here’s where it was cited from, here’s the line that it was in, you know, something like that. So that you can say, all right, this is a solid jumping off point. I, at least now know what I’m looking at.
**
48:10**
Daniel Englebretson
You know, I’m trying to get a sense at this point of what would that jumping off point need to look like so that you don’t actually have to read all 100 documents, but you feel like you’re in the right direction, you know, to know what you’re looking at and where you’re trying to go with it. So you can save yourself actually reading 100 doc or at least making sense of 100 documents out of the gate. Is it what would have to be like? Even if you don’t know how that would be, what would it have to look like? What was. What would the machine have to tell you for you to feel like it? You can trust that.
**
48:50**
Simon Walmsley
Good question. I, I think maybe going back to what I was saying about the next stage of after submitting the quotation is when we have this technical Alignment phase with the customer. And they’ll all have different formats the way they do it. But invariably it’s, you know, we’ll. We’ll receive an Excel form sheet with, you know, 50100 questions of, you know, are you complying with this point? Do you comply with paragraph number blah, blah? Are your showers constructed from this material? Do you meet this require? And it’s almost like a comprehension exercise of, you know, have you checked this? Have you. You know, do you comply with this? Do you? And if at any point in that, this sort of compliance checklist, do you say no? Well, then you’re expected to raise that as a formal deviation.
**
49:45**
Simon Walmsley
And sometimes in their, maybe template or their sheet, their document for deviations. So sometimes the compliance can actually be driven by the customer. So we’re going to have lots and lots of history of ones that we’ve done in the past that we could maybe amalgamate and kind of preemptively sort of find some common questions that keep cropping up that we can then, you know, preemptively, you know, do our own compliance. Because the thing I like about this project is I’m pretty sure that my colleagues will all have a slightly different way of working, and they might be looking for slightly different things because it’s all based on their own individual experience.
**
50:30**
Simon Walmsley
I think, correct me if I’m wrong, but as you said, this, this sort of architecture that we create now on day one isn’t going to be exactly the same after many weeks or months of using it. And I imagine that the more we use it, the stronger and more robust it’s going to because we’re going to be adding in things for it to look at.
**
50:56**
Daniel Englebretson
That’s what you ultimately want to have. Yes. And so we have to engineer it in a way that’s possible. That’s what you ultimately want.
**
51:03**
Simon Walmsley
Yeah. And we want to, I guess, avoid Groundhog Day, that we’re not asking ourselves the same questions over and over. If we can collate these into a sort of central repository, it’s, you know, for all the hundreds and thousands of quotations we’ve put together as a business, there’s got to be some common groundwork here that can save a lot of time.
**
51:28**
Daniel Englebretson
Yes, yes, exactly right. Yes.
**
51:31**
Simon Walmsley
And also as well, whenever we’ve had issues, say that a quality issue or discrepancy. Steve was hinting at something at the start of the call. We had a particular problem with the coating material on some bolts. And that piece of information was Buried in a specification that was linked and linked away. And you know, sure enough it was there. And you know, it’s our bad as a business because we said, yes, we comply to all of these points. And it caught us out and it cost us money as a discrepancy. So when we talk about that’s a lesson learned. Well, I think something like this by documenting it so that next time we have an rfq because we know we’ve been burnt on an issue of bolting specification or material we could specifically ask to look for.
**
52:21**
Simon Walmsley
Is there any, you know, we’re not just sleepwalking into the same issues. And by adding and building to it’s just going to get stronger and stronger that our compliance is there.
**
52:35**
Daniel Englebretson
Okay, yes, that is several things you said are dead on. And going back to how did you refer to it? You called it. Well, I don’t remember what the word used, but specifically not reinventing the wheel. One way I like to visualize this is, you know, the 20 questions asked. You always got to ask these 20 questions, but you’re tired, so you forgot to ask question 17. How can you keep that from happening? But as an organization, your organization knows the 20 questions, but no one person knows all 20 questions. And so how do you get all of the people knowing all the 20 questions every time? That’s the kind of thing we’re looking for. And those questions change depending on the scenario, and they change over time. And so it’s like, what are the questions?
**
53:21**
Daniel Englebretson
So that’s exactly the kind of thing we’re trying to surface in terms of what the engine is looking for as it’s doing the looking. But you still need the human to know, okay, here’s what this means, or you still need the human to learn. Oh, we’re starting to see this. We need to make sure we’re asking these questions because the machine can only do from what it’s seen before. It can’t just magically come up with something new, or at least not truly new. So as the regulatory environment changes or as the quality environment changes or as the customer product line changes, you know those things. And we have to make sure that the machine can reflect that change in reality. So, so we’re totally on the same page. So we’re in the last five minutes here.
**
54:03**
Daniel Englebretson
So now that we’ve kind of gone through this, the question I’ll put back to you for our final five minutes here is there anyone else on the team that we would want to ask a couple of questions of. Before we take the first pass at this, to make sure that we kind of thought about this well enough. Is there any. Can you think of anybody we would want to ask? Hey, what do you think would add the most value? Or what sometimes gets lost or what do you think customers would value most? If we could do this streamliner accelerate thing, Is there anything that jumps out at you that we would want to collect from other people related to the process before we kind of hammer out the first wrap, basically?
**
54:44**
Simon Walmsley
I don’t know what you think, Steve, but I’m wondering whether Matt should perhaps get his insight, because I know one of the questions you said at the start, Daniel, was what creates value for the customer? So Matt’s our head of sales. He’s the department head. So getting his insight, I think would be helpful.
**
55:03**
Stephen
James? Yep. James How? Engineering manager. I can send you these names or Simon can.
**
55:12**
Daniel Englebretson
Okay, okay. Yes. And so as you’re thinking about those people and those questions, I love like, yes, it’s about customer value, but it’s also now think about competitive differentiation. How could the fact that you can move faster or the fact that you could be more thorough, how could it help you win more deals, or how could it help you differentiate in the market? Like people who really understand customer value and they start to see how the equation is shifting. Oh, if we could respond in 24 hours to this hundred page thing that would do all of this stuff, you know, so those kinds of. It might not be possible, but it’s like, what could we aim for? You know, that’s also what I’m thinking about.
**
55:52**
Daniel Englebretson
So what I can do is take this conversation through it and come back to you and say, here’s what I think we would want to ask. I’m going to try to keep it to like two or three questions. And if you guys have others that you want to ask, that’s cool too. And then the hope is that over the next day or two, we can hear back from anyone who you want to ask that of. So that can just be included. And the biggest thing I’m trying to avoid is building a system where the result is garbage because we just forgot to think about this one thing. And so I don’t think that’s happening here because it seems like we have a pretty good look at it.
**
56:26**
Daniel Englebretson
But the last thing I want to do is run everybody throughout a build a thing and we just totally miss this thing over here, that this kills it, you know, so that’s what we’re trying to avoid. Cool. Well, I really appreciate the time, energy going through this. It’s. It’s very helpful. I’m glad we’re on the right track. B.
**
56:46**
Bhumika
What would be the. Yeah. What would be the next steps, Daniel, to do we meet.
**
56:50**
Daniel Englebretson
Yes.
**
56:51**
Bhumika
Next week for the. Or maybe this week for the, like, four hours that you were talking about.
**
56:57**
Daniel Englebretson
Yeah. So I’m gonna. I’m gonna come back to you today with a couple of. A couple of questions type thing. You guys can decide who and how you want to collect that. And then once we have that information back, I can do the draft of the flow. And then when we do the training, we’ll. We’ll use that drafted flow as the example. So we’ll be going through that as the example so that people can get their arms around it. So, so that. So once we have the data back, I’m probably needing. I can’t remember what I put into the schedule, but a day or two to, like, build the flow and materials and then we can have the workshop. So. So that’s the flow. And so once we have the workshop, we, everyone involved will see the flow unfold, be thinking about this.
**
57:37**
Daniel Englebretson
We’ll collect other ideas. We’ll leave that workshop. I will go build a second version after that workshop, and then we will come back for a final. Okay, here’s. Here’s what we built. Here’s how it works. Here’s how you can think about it and use it.
**
57:50**
Stephen
And then.
**
57:51**
Daniel Englebretson
And then that’s kind of where we end the piloting or of a prototype. So that’s how this is going to flow.
**
57:59**
Bhumika
All right, so wait for you for the questions then.
**
58:03**
Daniel Englebretson
Yeah. And I’ll do it this afternoon so that you can. We can make progress on this or. Yeah, I’ll do it this afternoon.
**
58:09**
Bhumika
Okay.
**
58:10**
Stephen
Amazing.
**
58:11**
Bhumika
I think Simon and Steve, you will have to answer those questions.
**
58:15**
Stephen
No problem. It’s exciting stuff. Thank you, everybody.
**
58:18**
Daniel Englebretson
Yeah, thank you. Bye.
**
58:20**
Bhumika
Thank you.
**
58:21**
Simon Walmsley
Thanks, Daniel.
**
58:21**
Bhumika
Bye.
**
58:22**
Daniel Englebretson
Bye.
**
58:22**
Simon Walmsley
Thank you.
**
58:23**
Bhumika
Bye.