About what I was hoping to chat to really interesting in your kind of engineering perspective, and would love to hear a bit about you and Fieldwire as well. So Hugh o'ford and bred in Dublin have lived in a whole bunch of different places in the US and Europe. Back in my hometown, have been in tech for pretty much my whole career. I started and sold a business before back kind of 2017 2018 I started a business. It was an AI audio co pilot for blue collar workers. We pivoted a couple of times, sold predominantly to customers in healthcare and hospitality and manufacturing. We exited that business in 2021 which was fantastic. And myself and my team are now back on to the racetrack for a second time to start our next business. What we're really focusing on now as the first use case is born out of a deep frustration and challenge that we had as we grew our last business, which was trying to keep on top of all of the various sources of unstructured data, shall we say so. A really good example is all the customer interactions, all the calls, all the emails, all the support tickets and how that sort of data gets used as input into decisions. Okay? And so that's, that's kind of where we're headed, and where I wanted to kind of kind of center the discussion today. But yeah, so that's that's been about me. Maybe you could tell me a little bit about you and and we go from there, sure, as you may hear, maybe on my accent, I'm originally from Germany. I've been here in the US now for like, close to 16 years.
And yeah, work at different, different companies, small, large, various engineering teams. So I would describe myself definitely as a general. It's not necessarily like one aerial focus. So my I had some excursions also into IP, like early on in my career, and it seems to be following me around. I work that, like at several startups, IoT startups, or I can't structure data. Was there, like, also, like, one, one big challenge, like at a pharaoh, the and I can go a little bit more into that. At Verizon worked on some FinTech products, a Verizon Visa card, it sort of works card. It's it's pretty big. And then we did also like a product called family money that got unfortunately shut down. It was basically like a prepaid debit card to teach kids how to call to responsibly use money with like, parental controls and everything, and so yeah, and yeah. Then I was at Etsy. It was more like an internal, internal support engineering team, and which also took me to Dublin at some point, because they, they brought, they built a new office. Yeah, right behind that, I forgot now the name, but behind the castle, basically, yeah, Dublin Castle,
yes. Easy to remember the name, yeah.
And so I lost her for about a week. And yeah, and here at Fieldwire, I'm managing the platform team, so we don't really deal with the consumption of unstructured data, and pretty much all data that we gather is here, fairly, fairly structured, simply because of the nature of the business.
How so is it is that, including sales conversations or prospects?
No, just the data that customers generate, right? Given that it's like in the in that industry, everything is very, very much to spec. Yeah, no pun intended, because that's what they need to do. That being said, there is certainly challenges around what the platform offers, right? Like customers can in real claim update plans and PDFs.
So for the construction sites, yes,
yeah. And so that data is can be considered, I think, like unstructured, right? Because we don't, we don't really know what, what customers like, really input there, how they had obtained it,
yeah. So your guys's customers are construction firms, builders?
Yes, yeah. So it could be anything from like the small handyman you know that that just wants to manage their their different projects, or different customers, one of jobs, up to like, the construction companies that manage like, building like, literally building like, a new building from scratch, right? Like those for all the all the phases, like, from the planning to the to the building specs, right from,
yeah, got it. So what I'd be interested in hearing your perspective on and the challenges associated with is, how does Fieldwire and you manage the fact that sales conversations, be it by email, be it By recorded call, support issues, be it by email or call or chat. Research Reports, how does all that data kind of filter through into, say, the top of the organization and filter down through, probably, through sales, through CS, through product to engineering, and what are some of the challenges associated with that?
The I mean from the sales side right there is the you have your typical CRM setup you have, then the all the other CRM adjacent, analytics tooling, right that you have life with social media integrations and so on. These things are funneled through, like Salesforce and like like, other integrations, so they end up eventually, eventually in a data lake on our end, from the support perspective, that is not directly connected to that. So that is like a separate track. And it would like you said the similar there as well. It's, it's just coming in for Zendesk. That's basically the, you know, the tool to manage customer support requests, and then the rest, how do you connect, right? Like support, sales and engineering is literally with people, right, just that they are trained and know when to like people and process so that the processes are established right when, when the sales conversations, for example, reach a certain point. Then there are certain things that engineering needs to do, for example, for certain trial setups and and sales has their own books to like, go through those things, right? It depends on how much we accommodate. There also depends on the size of the contract, etc, right? Because it's all sure effort. Yeah,
so And tell me about the nuts and bolts of that. So something comes in by email through Salesforce or email through Zendesk or whatever. What happens with that. And do you ever see it?
Not from an engineering standpoint, all that is is managed by those you know, same tools like Salesforce,
so you don't have PMS that share that with you. Um,
they do, but it is already very curated. So filth wire has a fairly mature business, so it's a very it's only being surfaced, and basically, I guess, on a need to know basis, because so I would say a lot, most of these requests are being handled in before they end up in engineering,
interesting, yeah, are there any problems associated with that? Do you do you guys as the engineering function wish things were a different way, or does it work? Okay, um,
I think the reality is, and this is like, really, not even like, connected to this business. But what I have seen in my experience is that the engineers usually want to be left alone, and the reason why is not because they are, you know, we like to sit in a in a dark room with your SM lights and just slit in the door for the pizza. That's not the case, but because there are way too many other problems to solve with competing priorities and the and that ranges from, you know, the feature roadmap, right, that that needs to, needs to move forward, right, the technical debt, bug fixes, and then just like, regular support. So these are the four buckets where I would say, like engineering, all our, all engineering organizations spend their time on that. And even if, like at large companies, like even at Verizon the, the even the most disconnected engineering teams those deal with those four maybe the they are weighted. These four categories are weighted differently, right? But there is no engineering team out there that just works on features that that may, may, may happen while you are in the startup, like stealth mode phase, but as soon as you move into like the more public phase, you're already starting to deal with tech debt, competing priorities to you know what features to work On? You know what? What would make the most impact, etc,
so how do you guys make those decisions? If the context of the of a lot of the intelligence in the business, yeah, from a data perspective, is not reaching in.
I think that's where the where the product managers come in that will have to connect with, I presume, presumably, sales and, or and, or support to give some of these priorities, and then, as the engineering manager, you push, you know, you negotiate, um,
and when, when they're negotiating with you, what are they bringing to the table as as leverage or as ways to negotiate? Are they,
yeah, I'm trying to think here like of a concrete example.
I think some of that is also market like, some factors are driven by market research, right? What the What the competitors stack up, right?
So they come and say, Hey, this big competitor has already done this, and we haven't built it yet, and we're losing deals and,
yeah, and that's that's like, more on the business side to also, like, know where you, where you you can be the old singing, all dancing, right, successful businesses, they know exactly where they're you've probably noticed, like, what is your not niche, niche, but like, what is your the space that you carve out where you want to be? Like, really good. Yeah. And, and so field wire has done the same. So there is intentionally some, some things that we that we don't, don't do, because that's that's just not our, our area of that we want to expand into. Yeah. That being said, I think that's a so competitive analysis, and then also looking at customer feedback, right? That that comes, that comes, usually through sales, actually not necessarily support, okay, simply because of the the account management perspective, right? That's, that's where we would usually get the most, the most, the most feedback from, is from customers who through their account managers in the sales teams. Then necessarily support. Of course, there are then support. There is an avenue of support where they try to solve a problem, and it will then, usually support and sales are working fairly closely together on on understanding this, which will then eventually can come to to product. Now I'm not private all those conversations that they have on how they how they do the prioritization.
But when it these features come to engineering, I would describe it maybe it's a wish list of priorities.
Yeah. And you guys can decide. You guys then work through and say, Okay, what's possible and what timeframe with what capacity?
Yeah, yes, technically, like negotiating on, like, the scope, right? Yeah, what does the roadmap for that look like? Yeah, and, like a, like an, and you notice, right? Like, like, we have mobile apps, we have back ends, right? And you want to keep those forms, you know, in sync features. It's not always possible, but at least that's the goal. So and that will need to be negotiated against tech debt. And I think that's the that's the biggest issue always, is because there is certain tech debt that becomes a blocker for development, right?
So in all the places that you've worked? Yeah, when you think of all of the unstructured data that exists in a business, whether it's in customer interactions, whether it exists in documents like, what are some of the good and some of the bad ways you've seen that being used? Or do you have, or do you have a wish? Or would you, do you wish that something could be done in a different way or in a better way?
So one thing that I've seen it where it was not done so well was, for example, advertising, product management team there, which was great was that they spend a lot of effort in getting in research before they went into development. However, they tended then to over index on that right? It's like, if you think of this product, you know, what would you what would you like to see? I mean, I'm oversimplifying here, but these were this because they tried to figure out, like it was a brand new FinTech product, right? They tried to figure out what parents care about for like this prepaid card for their children. And the issue there was that it was almost taken in verbien to the to the engineering teams, without, without really unders, like drilling more into what, what certain features mean. So that's what I what I thought it would have been good to break some of these features out. Have more discussions around it, like, really, really dig into them more on the drawing board before, before starting building them interesting.
So use so have a deeper understanding of the ideas before they get,
yeah, deeper understanding of the ideas, right? Because they they, the research that they did was basically just asking people, like a Greenfield, you know, would you like this feature? How would you rate that? Where I've seen it work, actually horribly. Was at life was basically talking to potential customers and taking their feedback, also verbatim, and then making a priority. So every week was a new feature, or the old one was not even implemented. So that's like this, like the shiny new thing that that every week changed, like that was not successful. Like you had half baked features that did not work. Where I've seen it work actually fairly well. Was at a Ferro where the feedback from the customers was also taken with a crane of salt. So there was then, like, some more deeper conversations happening with them about like, but do you really need it? That also means, however, that was possible, because the team there had a pretty good understanding of what the space that the customer is in, right and we're also challenging some of the things, how they looked at IoT, how devices may should behave. So it became more of like a conversation versus a just like here I take input and try to turn it around around one to one. That was probably the most successful that I've seen as a pharaoh was that, yeah, the also pushing back on the customers to say, Listen, we will hear what, what you need. But here are, here is a more general approach, you know, that would also solve, like other problems for you, yeah, you know, you may have to change maybe some some things how you're looking at,
yeah, interesting, interesting. And that
was also done in combination, like with sales, and the sales engineer came in, right? That was that helped a lot with, like, the that was adjacent to the to the actual salesperson, right?
So we have about three or four minutes left, so I want to just tell you a little bit where I think we're going to go, but we're, we're so early. But I mean, you know, the startup world so and just to, just to see what you think. I mean, one of the things that we're really trying to focus on is that the that the best source of of business intelligence for businesses lies in their unstructured data. But leaders like you, and leaders in products and leaders in commercial can't access it all right? So it's not possible to listen to every call or to see every customer conversation, to read every document, and if you think of the volume of a lot of those as you get to your stage, I mean, you know, the hours of calls every day, the 1000s of emails, 1000s of documents around and being created. It's an enormous quantity of data that seems underutilized, undervalued, under, leveraged to make better decisions. And if you think about the world of business intelligence, it's kind of interesting, right? Because we've got companies globally spending by $33 billion on structured data, yeah, and it represents 20% of a company's data. And then when you look at unstructured data, there's nothing, right? It's like manual processes, combing through it, tagging it, moving into another system, copying and pasting it, and it's just a really slow, manual process. So what we're thinking is, Okay, what if we connect it to all these unstructured data sources, and based on certain use cases or certain jobs to be done, we could produce business intelligence out of that data. So, for example, all interactions. So I don't know if your CEO asks questions of your product colleagues like, hey, what's all the positive feedback and negative feedback we've received on the latest release? Or maybe CEO asks sales leaders like, hey, what which sales positioning is working best? Or, you know, maybe they come to engineering, they say, Hey, what are the performance issues that we're having, and how are customers responding to that? Right? And these are the sorts of kind of business critical questions that just take a really long time to answer. So we could connect to the data sources. We could understand what each person in a business cares about, and then we could deliver to then curated intelligence on those certain topics, starting super simply with an email report or a Slack report delivered every day, every week, whatever, whatever is appropriate, and then over time, building up more types of outputs. So you could imagine that 50 hours of sales calls condensed to a three minute podcast, 20 hours of sales videos condensed to a couple of minute of a video reel that you can watch and and and try to bring a lot of that unstructured data alive so that people can see it, listen to it, consume it, read it, whatever the case may be, and actually start to use that that's that's the area, the direction that I think we're moving in. And so I'm just curious, given all your experience at so many places, what are your thoughts? Are positives and negatives? Um,
um, yeah, I think you're talking here about, like, large language models that that would work for the data, and trying to, like, figure out what the the, what is the quintessence of this unstructured data based on the question that was being asked. There are, there is a challenge, certainly with unstructured data just actually finding information, not even like synthesizing it, but just finding it right. And there are companies in this space like lean and dash that that try to solve that, yeah, but it sounds, from what you are describing, that you want to go a step further and actually like, also provided like more like like summaries of the
insights. Yeah, the value with the insights, the insights, yeah.
I do think that the data would be valuable. The challenge, I think will be, will be definitely in the training of the llms, right? That that these models, while they are, I would say, pretty ubiquitous in the meantime, right? They still tend to, like hallucinate. They still tend to, they can be influenced, right, oftentimes, not in a good way. So you basically, you keep asking questions, and then eventually it will tell you what you want to hear, what it thinks you want to hear, right? So, but those are, those are, I guess, like more technical challenges than the than actually absorbing the data and providing that is that did i Yeah,
you captured it exactly. So, last question from my side, imagine you could get a report to Slack or to your email every Friday, and it contained insights from the unstructured data that that exists within your organization, right? So anything that comes into field where we could generate a report on it. What do you think that report would show? Where's your blind spot? What's your black box?
I think, from an engineering standpoint, I would, I would want to know if there were any performance issues that customers were talking about. Maybe they just mentioned they did, like in one sentence, oh, this is slow, or this did not work. Yeah.
So performance.
So again, this is just you asked what I would look for, right, yeah, but
that you look after the platform, right? So that, so it makes sense, yeah, yes.
I think maybe from like, as a product manager, they would want to know, like, what is the what is like a feature like requests,
yeah, relative to the part of the product. Reason
about like functionality, maybe that's that's a better,
yeah, interesting. We're super early. We're playing around with some stuff, with some design partners where we're testing it out, these conversations are just insanely valuable. I was so pumped when you replied, Because I don't speak to many folks, as you describe yourself, of engineering generalists, I speak to many more commercial and product people, and so that's why this perspective is super interesting. I'd love to keep you up to date on what whatever we end up doing. Maybe
it'll be something totally different. Who knows? You know how this world
goes, but, yeah, but thank you so much for your time. I heard I really, really appreciate it. Lovely.
This was very interesting conversation.
Yeah, yeah. It was great to meet you and keep you posted. And if you ever back in Dublin, although field wire don't have an office here, but if they ever do, and you're down behind the castle, let me know. Go meet for a beer. Yes,
yes. Nice, yeah. My, my, my peers there took me, took me through, like, downtown to, like, a few pubs and but we ended up, always ended up at grogans, not sure if you're that one. Yeah, too,
too familiar with grogans. I was there on Friday. Yeah, did you? Did you get the toasted sandwiches? 100%
to get the toasted sandwich,
yeah, if you saw how they were made and where they were made, you wouldn't eat them. It's not, I don't think they would score very high on food hygiene. But it doesn't matter. When you're consuming alcohol, it kills all the children.
It's all getting like, cleansed.
Where in the US are you?
I'm in San Jose, San Jose, California.
Yeah, beautiful,
beautiful part of the world. Great weather all year round, and nice beach life, yeah,
yeah. Although this this week, it's a little bit more foggy, which is also good, because we had, like, a lot of lot of heat then the last few weeks, last few months. So,
great, great. Well, her art, so good to meet you. Hopefully we get to meet in person at some point, at some stage, but I'll keep you posted. And like Guys, let's stay in touch. Indeed, You
always reaching out.
Take care. Bye, Bye.
I Don't Know
music.
Here's a short video to show you what is possible with centric. Centric is everything your customers have said, sent or suggested this AI generated video was automatically created to recap a number of calls we've had over the last few weeks. It's discussions with VCs, but could be any recorded calls you have with customers or prospects. The snippets and feedback you hear were created by centric AI to distill 300 minutes of conversation into the most important part for the purposes of confidentiality, names, funds and other identifying characteristics have been changed or removed, but their voices and opinions are real, which we hope you enjoy. We started by asking what it takes to raise a 1.5 to $2 million precede in today's market. Here's Oliver from solo fund raising
1.5 to two as a pre seed is like, you kind of just need to go to credible energy, but you just literally cut the bottom right, yes, the company in The future. It's kind of A waste of Time. I
