Hiten Patel: Thank you for joining us on today's episode of the Innovators Exchange. Today I'm co-hosting with my colleague, Archie Stebbings, who leads our work with the risk and compliance solution providers, and we are delighted to have with us today, David Benigson, the founder and CEO of Signal AI. Welcome, David.
David Benigson: Hey there! Thank you for having me.
Hiten: Why don't we start things with you giving a brief intro to your role in the company that you lead today?
David: Brilliant. Well, yeah, I'm David Benigson. I'm the founder and CEO of Signal AI. I started the business 11 years ago now, in my parents’ garage, of course, where all good AI startups begin, with my co-founder, who's an academic Dr Miguel Martinez. He was finishing his PhD at the time that we started the business. And we began Signal AI, really under the premise of observing the world becoming more volatile, more noisy, more complex. The number of risks and issues that large enterprise businesses facing off against had sort of never been harder and more complex to navigate, and we observed that these businesses lack the sort of external intelligence and data to be more predictive and preemptive, to get ahead of those risks in a more effective and quantifiable fashion. And so we've spent the last 11 years building Signal AI as a platform that can aggregate the world's unstructured external data and apply machine learning and AI to extract these signals of risk and reputation intelligence from the data, and then deliver that to our clients so that they can make more confident decisions to navigate these complex times.
Hiten: Awesome, and I guess, in layman's terms, for some of our listeners kind of just walk us through like who an example user of Signal AI may be, and the use case in which it's been deployed.
David: Yeah, sure. So, we work with about 40% of the Fortune 500. So, it's typically large multinational enterprise customers, you know, think about the world's biggest banks, or the world's biggest brands, the world's biggest tech companies, pharma and healthcare organizations. And we're typically selling into a couple of different functions within those organizations, the corporate risk function, the enterprise risk management team, corporate communications, and then a little bit into the sort of first-line functional risk areas as well, could be the head of supply chain, or the head of ESG, the head of regulation and policy affairs, for example.
And really, what we're arming them with is this sort of external risk intelligence, this ability to look outside of their organization and use this data and technology to be more predictive, to be more sort of quantitative in their approach, to spotting emerging trends, emerging threats, emerging risks, and issues, and reputational challenges to their business. And we're doing that by essentially aggregating this breadth and diversity of data that sits outside of their business from licensed premium media, to the world of social media, to regulatory and policy data, earnings transcripts, broadcast, TV, radio, podcast, data.
And then we're using our AI machine learning technology to, in real-time, sort of extract these emerging signals from the data. That could be the changing sentiment, that could be risk events, that could be supply chain disruptions, changing regulation and policy. And we're delivering that to our clients in a set of alerts, in a set of dashboards and visualizations. And now, increasingly, with this sort of agentic workflow where our clients can ask natural language questions about our product and our data and get back natural language answers. So, similar to a ChatGPT or a Perplexity AI. But sort of sitting on top of this really high-quality set of data and sort of retrieved and extracted through these very specific AI models that we've developed over the last decade.
Hiten: And before people had a solution like yours, how would they manage what you just described?
David: Gosh! Well, the world of sort of monitoring the outside world as an industry has been around for a long time. I think probably we were one of the first organizations to bring this much more sort of AI-led sort of technology, real-time approach to that opportunity or that sort of value proposition. I think historically organizations were doing this in a very manual sort of ad hoc fashion. So, if you speak to most risk teams today, they'll do an annual risk assessment as an example. But it's sort of quite backward-looking, it's a moment in time. They may do it at the end of the year when they're looking at their risk register, and they're trying to assess their risk exposure.
But that is very manual, and it's very ad hoc. They may be working with advisors or consultants, but I think in a world that is so dynamic today, so volatile, moving at such a pace, they need a much more proactive approach to sort of getting ahead of risks and issues. They need to be able to respond in real-time, and they need to be able to sort of track and understand the conversation and the narrative as it flows through these different, sort of, media channels, as it flows through these different stakeholder groups. And so they need access to this data in a much more real-time and much more dynamic sort of proposition.
Hiten: Gotcha. So look, thanks for painting the picture of where you are now. I always also love to delve into the David Benigson backstory. You aren't often just born out here at the point in time now, so talk to us a little bit about kind of the start of your own career before Signal AI. Talk to us about what the outset looked like, and how the journey played out.
David: Sure, happy to do so. Yeah, well, I was pretty young when I founded Signal AI. I was in my mid-twenties; I was about 25 when I started the business. As I said, I began it, I found it in my parents’ garage, and my parents are both entrepreneurs. They've run together an executive search firm for over 35 years. They emigrated from South Africa and came to the UK with nothing and started this business, I think, around their kitchen table, not their garage. So, I sort of was brought up in an entrepreneurial household. The idea of sort of starting your own business and building something and having ownership was sort of very native to the way we thought about things. Initially, I wasn't quite sure what I wanted to do with my career. I'd studied English literature.
I then went and did a law degree. I started a career in law and quite quickly realized that wasn't going to be the right pathway for me, and that wasn't going to be the right environment for me. And I had this sort of entrepreneurial hunger. But for a couple of years prior to founding Signal, I had the opportunity to actually work for Jamie Oliver, the chef and entrepreneur. He was an amazing individual, still is, and was right at the peak of this sort of entrepreneurial journey he'd been on, building out his organization, and so I sort of got an opportunity to sort of see firsthand another entrepreneur at work in a very, very different industry. But someone who was extremely exciting, extremely creative, just very natural in his approach to how he was building his business, you know, certainly hadn't come through a sort of corporate background or a structured business background, and just really had this very ambitious vision for what he wanted to build and what he wanted to do through food and through cooking. So that sort of, you know, that set of kernels of different sort of inputs really led me to want to start my own business and found Signal AI.
And it was a sort of random confluence of different events that ultimately sort of led me to the business. But one of them, foundationally, was seeing sort of firsthand organizations struggle with the volume and velocity of data and information, a number of different examples of businesses being caught off guard, caught by surprise. And I met a professor who was an expert in machine learning and AI, Dr Udo Kruschwitz, and he told me about this new government initiative called Innovate UK, where they were funding academic research and bringing it into a commercial context. And I described, sort of outlined the straw man for the idea for Signal, and he said, well, why don't we apply for an Innovate UK grant together? So we applied for this grant. I still had a job at the time, and a few months later we were awarded, I think, a £250,000 grant to be able to take this academic research in machine learning and AI and apply that into a commercial context. And as part of that grant, I needed to find an academic who was finishing their PhD to sort of apply their research into this world. And so Udo introduced me to Dr Miguel. He then sort of joined me in the garage, and that was how we sort of set off on the journey of building Signal AI.
Hiten: That's awesome. I don't think we promote enough of these stories. You're not the first founder who's had a similar story. Actually, we had Christian Nentwich from Duco on recently, and he was working with an academic partnership. And if you look at everything we're wrestling within the UK about where can we get innovation, where can we get growth? We have these great education sectors, hearing stories like yours that are born out of marrying private sector challenges with kind of all the expertise in the education sector. It's really warming, and I'd be passionate to bang the drum more on. So it's, thanks for sharing that, David.
David: Well, we were very lucky, if you think about 10, 11 years ago, sort of the confluence of different things that were happening in the UK Tech scene at the time. Certainly, there was a very pro-innovation government, introducing a number of these different initiatives to help support the infrastructure behind technology, whether that was Innovate UK, R&D [Research and Development] tax credits, SEIS [Seed Enterprise Investment Scheme] and EIS [Enterprise Investment Scheme] investing, and a whole bunch of other initiatives that were really being funneled into that tech ecosystem, the silicon roundabout culture. That was number one. Number two, you had sort of the emergence of a whole bunch of early-stage venture capital funds coming to market.
When I first found Signal, and we were raising our seed round, there must have been half a dozen seed funds, maybe a dozen seed funds maximum, in London. I think there is now, you know, over a 100, probably of seed funds and established sort of angel networks and syndicates, etc. So there was this access to capital that was emerging in the scene. And then, third, as you say, there was a whole bunch of incredible work being done at the academic level, and the UK still is one of the, you know, the highest quality academic technology ecosystems, certainly in the world. And alongside us were organizations like DeepMind being founded, and many other of these deep tech AI businesses. And so there was this whole sort of burgeoning excitement and set of talent that was coming out of universities at the time. So we by chance, or I by chance was sat at the center of those three trends, and very luckily was able to sort of ride some of that.
Hiten: Yeah, it's great to hear that some of those initiatives and policies work, right? We're just forever drowning in bad news and too many short cycles of expecting overnight successes. So hearing you talk, and it's a theme that's been resonated actually across the show. I think Sherry Coutu talks about her research for the government in this space. Charlie Kerr, who founded a company, was successful similarly. So look, you're in good company, and I think it's great to hear. One more question for me on this point. Like hearing you talk about the history of your journey, I guess again there's a common theme that you were into AI machine learning before this whole ChatGPT era, like many years before. Just give me your perspectives on how helpful, how much of a hindrance is it that you've now got what feels like a gold rush and a lot of people turning up more recently. Right? Given, you've been at this for 10 years. Just some reflections on, kind of, from someone who has outlived the most recent kind of spike in interest.
David: Yes, no, I think Net-Net, it's hugely positive for Signal AI, and for my business, you know. First and foremost, a huge debt of gratitude to Sam Altman and OpenAI, because my mom finally understands what it is that I do.
David: So, it's sort of broken through into consumer consciousness into mainstream consciousness. And I think that's very powerful. Probably for the first six, seven, eight years of building Signal AI, particularly selling into these large global enterprise businesses, we had to do a lot of convincing around why having an AI strategy was important, why doing things differently was going to become a necessity and the benefits and value of that. And at times we were met with a certain degree of cynicism and skepticism by the client's side. Well, in the last sort of two, or three years, that has been flipped on its head very dramatically, and almost every organization now knows they need to have an AI strategy. It's coming very much top-down from the most senior levels within their organizations, often from a board level. And so, Signal is a sort of ready-made solution for that.
We've been around for over 10 years. We're established as a brand, you know. We apply things. You know, we apply our products and our propositions in a very compliant and enterprise-friendly way. And so for organizations who are now proactively looking for an AI strategy, we're a great solution because they may have already been working with us, or they want to now work with us, and they want to do things in a different way. And then, at a more sort of technical level, there is a phenomenal opportunity to combine these different technologies together. I think in the last couple of years everyone's become very fixated on generative AI and large language models. We often remind our clients that it is just one technology in a whole family of different AI and machine learning approaches. And it's very important not to think about it as a sort of elixir for all problems or all parts of the workflow.
We are experts at Signal and have 10 years of expertise in what we call discriminative AI. The ability to train models, to be able to discriminate and retrieve highly accurate signals and information from large data sets. And so, we're very, very good at finding the signal from the noise. Generative AI is profoundly incredible at being able to generate new information, synthesize data, write in a natural language format, respond to natural language questions. We believe the combination of these two technologies, combining discriminative AI with generative AI, is enabling our clients to get the best possible value from our products. So we use our discriminative AI to first retrieve and extract these high-quality signals and make sure they are highly relevant, highly accurate, truthful, factual, and then we can use generative AI to summarize those findings and synthesize those findings and deliver that proposition in a much more natural language sort of conversational experience. So, for us, it's been a ‘1+1=10 scenario’ to combine these different technologies.
Hiten: Awesome. Just going back to your opening statement. I don't think my mom still knows what I do. So, I need a Sam Altman like you've had. I'm going to throw the mic over to Archie now and invite him to kind of delve into the market you guys are playing into. So, over to you, Arch.
Archie Stebbings: Thanks, Hiten. And David picking up on your last set of comments around the combination of discriminative and generative AI. I guess when we're in the market speaking to clients, speaking to organizations who are buying solutions like yours, or like those other ones on offer. We're still coming across very divergent views on whether the time is right, whether solutions out there are a good fit. I guess, how much of your time are you spending around the human coaching, bringing people on the journey with your element of this versus, you know, taking a product in a box and standing up at the door with a price and a narrative.
David: Yeah, it's a great question. I think it's a good mixture of both. I mean, historically, at Signal AI, we would have sold into the organization sort of. And I would describe that as bottom up. And what I mean by that is we're finding specific functions, specific business units, if you like, who have a specific problem they're trying to solve. And we're selling our product solution very specifically to help them solve that problem. And the fact that we're using AI is a sort of, in the background, almost, it's a sort of how we make the sausage sort of scenario. And yeah, it's important and helps us differentiate and maybe helps us deliver value that others can't. But first and foremost, it's figuring out the customer problem and then bringing the product and technology in to support that.
What's interesting about the current market context is, we're still doing that. And that's still a significant part of the way that we go to market. But there's also this sort of top-down selling as well now, where the CEO or the board have made a specific sort of animation towards, we need to use AI, or we need an AI strategy. And they're almost looking for solutions or problems. So they're thinking technology first and almost problem second in an interesting way. Now, what's really interesting about that sort of go-to-market paradigm is you're often coming in at a much more senior level within the organization. And so you're able to have these very strategic conversations and then figure out how you can plug in your technology and plug in your solution to help solve real business problems.
But what I think we've also found in the market is a bunch of experimentation, a bunch of ideation, that in some cases hasn't then led to real business value, and I think that's quite dangerous and risky. And so we may see a little dip in the passion and froth around AI as organizations figure out how to sort of separate the wheat from the chaff of yeah, this was an interesting experiment, but it never had any real business value. And I think we've got to go back to what are the real problems we're solving. And then how do we use the right technology to solve those problems if that makes sense.
Archie: And do you see any differences in, I guess the use case or the product element that you're selling? Where, for example, I think you mentioned risk solutions. You mentioned compliance-related solutions. When the implications of quote unquote, the dreaded word ‘hallucinations’ are an issue around regulatory compliance, or a fine, or exposure to something you didn't see coming. Do you see any divergence in the way people are either grasping the opportunity to deploy this tech or being more cautious?
David: Yes, certainly. I mean, I think when it comes to risk management, when it comes to compliance. In fact, when it comes to using data for real decision intelligence. I think most large organizations have a relatively high and understandably so threshold when it comes to trusting these tools and ensuring quality and accuracy is central to the solution and the output. When we're approaching large organizations, we're already positioning our AI and our capabilities and thinking about three things. The data that we're sourcing and making sure that it is of the highest quality, and that it is also being sourced compliantly. We're definitely in the consumer world, seeing this thing play out where a lot of the large language models have hoovered up the world's information, ingested it into their systems and engines, and not necessarily paid the right royalty fees, or have the right licenses to compliantly access that data. So, Signal is compliant first, and the data and content that we ingest into our systems have been compliantly licensed. And that's part of our value proposition to large enterprise customers.
Number two, there is this issue of accuracy and quality. And this is where, again, we believe the rag approach we're taking of combining discriminative AI with generative AI enables us to mitigate and reduce the likelihood of hallucinations or misinformation dramatically. The other benefit you get is the ability to have an audit trail, a set of citations, links, and transparency as to what's driven the answers. So, when a client asks our system a question and gets back an answer, within that answer, all the links, all the citations of where that answer is being driven from. And I think those three things are very important to enterprise customers who are going to make use of this data for real decision intelligence. Has the data been sourced compliantly? Can we trust the quality and accuracy? Is it auditable, explainable, and transparent? What's driven the answer? And you sort of need to tick those three boxes if you're going to expect a senior decision maker, particularly in the area of risk and compliance to warrant using a system like this.
Archie: Fascinating. And that point around the LLMs [Large Language Models] putting their arms around all of that information that's out there, ideally in service of positive outcomes for those who are using them. When you think about the importance of unstructured data and content nowadays and the various different sources that that's emanating from whether it's social media, forums, podcasts, other dark web sources that previously were unexplorable or just the volume of stuff that was out there was so great people didn't know where to begin. I mean, where are you seeing the most potential there going forward? And there's a secondary point. How do you think about ensuring that the quality of insights emanating from those sources is appropriate?
David: Yeah, great question. So, I mean, our data sets. We started with licensed media. And that and that continues to be a cornerstone data set for us because of the quality within that data set, you know. So having access to that premium media and using it as a sort of tool for correlation from the other, you know, from the other signals that we're extracting from these more disparate and alternative data sets continues to be really important. So, if it's being reported in the FT, or the Wall Street Journal, or the Washington Post, we give it a higher weighting, if you like, than something that we've extracted from a random blog, or from a tweet, or even from a discord channel. Nonetheless, the sort of signal, the canvas from where we want to extract the signal from, is diversifying dramatically and is becoming increasingly critical. So, you know, with media, we're also accessing a vast array of different social media channels.
We've also moved into what I would describe as alt social. Areas like I just mentioned, Discord, Mastodon, and Truth Social. These are places that we're not necessarily spending much time on day-to-day but are becoming increasingly influential, you know. Maybe the most likely place you'd find your passwords have been hacked and leaked, or a protest is being organized outside of your offices during your AGM [Annual General Meeting]. And so, as an organization, you can't bury your head in the sand. You need to have radar into these areas and be able to pull out intelligence and insight as to what is emerging from those different platforms. Regulatory data and policy data, increasingly driving operational strategy and being able to sort of scan, and analyze the emerging changes within that landscape. Absolutely critical. You know, you mentioned multimodal data, TV, radio, podcasts. If we think about the US election last week, the influence that podcasts had as opposed to mainstream media was very profound. We're monitoring the 30,000 most popular podcast channels around the world. And I think that that as a channel will continue to grow and influence. So I don't think there's any one specific data set that outweighs the other.
I think it's actually about linking these very diverse data sets together, enabling our clients to have a perspective and look into every dark corner of these sort of unstructured worlds, but also be able to enable them to sort of traverse and understand how narratives and conversations and risk events and issues sort of spread through these different channels. You know, what may start on social media spreads into mainstream media, gets picked up by alt social, you know. A misinformation campaign starts there, that gets picked back up by mainstream media and propagated, which leads to, I don't know, some politician speaking about changing policy. And you want to be able to let our clients sort of traverse naturally through those different channels and understand how those narratives and those issues are spreading.
We had a fascinating example with AB InBev, and you'll remember the huge fiasco where they asked a well-known trans rights activist to represent the brand. Now, that wasn't just an issue because there was then a backlash in regard to folks saying that that individual wasn't aligned with their brand. There were then human rights issues, there were security issues, there were some policy issues, there were reputational risk issues. And all of these things sort of spread over a number of months, quarters, and into years. And so we need to enable and empower our clients to be able to sort of monitor and track those stories and those narratives as they evolve and go through those different parts of the conversation cycle.
Hiten: Really powerful. What you lay out there, David, just coming in on when you describe that the picture that's built in my head was this trade-off between the level of trust or factual accuracy on your data source, balanced with the forward-looking and the backward-looking, like the examples you give. It feels like a lot more of that forward-looking, predictive things may live in areas that historically don't have a heavy level of trust, probably because of the editorial and reporting requirements. But you probably do need to access them to predict what's coming. And it's really interesting. The pitch you did.
David: And that, I would say, is one of the big next frontiers. You mentioned the word predictive. But it's moving from what I would describe as descriptive analytics, i.e. being able to use AI and machine learning to describe what's happening within the data and explain it with a high degree of accuracy, to then being able to move into describing what's causing those trends and those events and being able to help our clients understand the clusters of conversations, the narratives, the storylines as they're emerging. And then the next sort of frontier is being able to be more predictive and being able to describe to the client what's at stake, and being able to analyze the blast radius of a particular risk issue or risk event, being able to predict or forecast events before they've occurred, being able to alert our clients to these sort of early warnings that things are bubbling up, and these unknown unknowns may be coming to impact their business.
A big thing that we do for clients is look at contagion throughout industries. So, if you're a big FMCG [Fast-Moving Consumer Goods] player in the market, an issue for one of your core competitors can very quickly become an issue for you. A conversation about talc in a totally different product area can become a reputational risk or an operational risk for you in a completely different use case. And so, how do you be able to sort of spot these emerging signals and get ahead of that risk of contagion before it spreads into your world.
Archie: David is that learning from precedent scenarios that are comparable? So you know the implication of a certain well-known podcaster endorsing a certain political candidate, or the AB InBev example you've just described, the patterns of social media chat that followed immediately after that campaign was launched. You can sort of spot the replication of those types of situations down the track and then be informed around, you know, what comes next, or the power of that from a predictive point of view.
David: Exactly. So, we've got over 10 years' worth of data now, where we've been extracting these sorts of signals, and we're able to decode and understand the risk events as they've occurred. We can connect all of those risk events and those signals into a knowledge graph. And we can understand, you know, what was the sort of the causal chain, if you like, from one risk event to another event. Once you've built out that sort of archive of data and you've connected all those causal chains up, you can start building more statistical models that become more predictive models that enable you to look at what the likelihood of one event occurring and then a set of events occurring thereafter. Now, we haven't quite gotten to that sort of simulation-type module yet in our product. We've just started launching our first sort of agentic type workflow, which is now live within our product, we can ask questions and get these sorts of answers back. But the future paradigm you can totally foresee is a client, a chief risk officer, a chief communications officer, maybe even a CEO, asking a question about a particular issue or an event, and the system providing a forecast or a simulation of what might occur based on these 10 years billions of data points within that graph and within that archive that we're analyzing.
Archie: Great. And we've obviously gone quite deep there on the various different use cases and the potential of what you guys are bringing to market here, and it's all very exciting, right? And there are a lot of use cases. You mentioned earlier that you're having conversations at the Board level, but also with very technical individuals, on very specific deployments of your product, within risk, or compliance, or supply chain, or whatever it may be. Can you do that everywhere? Or are you really reliant upon there being certain individuals within organizations who kind of get it and are willing to espouse what you do to that broader organization?
David: Yeah, great question. So we've certainly learned over time that narrowing your ICP [Ideal Customer Profile] down is absolutely critical and having a very crisp view of what that ideal customer profile is, and sort of continuing to refine it. And you know, when we were early days just out of the garage, trying to find that original beachhead. I remember reading ‘Crossing the Chasm’ and being very influenced by that as a sort of business strategy model, you know, we narrowed in on financial services, and we narrowed in on corporate communications leaders, and that was like our initial beachhead. And from there, we sort of branched out by industry, and then over time, we branched out by function, and then by time, we sort of branched out by geography. Today, the business is pretty industry agnostic. We sell into almost every sector in every industry.
Naturally, there are certain industries that are more exposed to risk and volatility, and reputational threat, but almost all businesses of a certain size and scale are grappling with a very complex environment today. And so, we're pretty industry agnostic. We are largely sort of domain subject matter agnostic. And we're language agnostic and geographically agnostic. But we are functionally specific. So, we're really trying to go after a number of very specific buyers within a large enterprise and solve their problems very specifically. So, the chief risk officer and the Enterprise risk management team may sit underneath them. They have a really difficult job. They are trying to manage and oversee risk across the entire organization holistically. They are not an expert necessarily in any one functional risk area, and their job is to make sure that they are protecting the organization writ large, that they own the risk register, and that they are reporting back on risk to the Board or to the CEO, or the CFO. That's a really difficult job to do.
Historically, that function has been pretty internally focused. Relatively small teams, given the size of some of these organizations that we're talking about, relatively small teams, and they've been very focused on internal risk management, and they've probably admittedly dropped the ball a little bit on being able to, you know, be external looking, and get ahead of external threats and issues. But that is where a significant amount of the risks come from. From the outside world into the organization, not from inside the organization. So we are specifically focused on that buyer persona if you like, and trying to provide that individual and that team with a much more comprehensive holistic view of risk to the outside world, to be more predictive, to be more quantitative, to be able to take their risk, register and apply it to all of those external data sets that I mentioned before, so that they can then bring that intelligence back into their organization, and have as much confidence and conviction about external risk management as they do on internal risk management.
The other side of that coin, you know, when a risk actually hits a business normally, that's when the corporate communications team is then brought in to sort of clean the mess up, if you like. But interestingly, although they are two different sides of the same coin, corporate coms speak a totally different language, use totally different tools, and sometimes don't even speak to the risk function until it's too late. And so our other core focus area is on that corporate communications leader and arming them with a much more high-quality set of data, a much more analytical approach. You know, comms, historically, has been much more intuition-based, creative, and critical thinking-based, not really data-driven as a function, not really framework-driven as a function. So how do we arm that function with a really high-quality set of data and intelligence and enable them to speak the same language as the corporate risk function. And so we're already selling into the corporate center and arming those two particular functions, hopefully with the highest quality data that enable them to look outside of their business.
Archie: And if I may, just as a final kind of point in question, the one thing we've always seen has been super challenging around those external risks, as you put it, is what's the impact? There's almost this amorphous sense of reputational risk that, in many organizations, there's a struggle even to place that on a risk taxonomy. Is it a risk? Is it an outcome? Is it an impact? How do we think about sizing it? I guess I get pretty excited about this agentic point because it's starting to help solve some of those questions when you think about precedents or predictive power around certain events.
David: Yes.
Archie: Are you seeing traction there? For that reason.
David: Totally. I mean, I think you know, reputation has always been one of the most important but unquantifiable, you know, value drivers in an organization, and we sort of know when it's been destroyed. But we don't really know how to measure it when it's there. So we're bringing a lot more confidence and quantitative data to being able to measure the impact on reputation risk and on reputation sort of impact when it goes awry. The other thing we're doing, which I think is very exciting, and this is another big frontier in the future for our business, and the industry as a whole, is starting to connect the unstructured world that we just talked through and described to the more structured world.
And so, can you start connecting all of these unstructured data points and data sets. And the analysis we do around sentiment, and topics, and risk events to more structured data sets around financial performance, around firmographic data. That is, of course, the Holy Grail. Because if you can start to tie back how you know reputation, how risk actually has a real tangible impact on business metrics then you've got something significantly impactful for the organization to make decisions upon. So we're just in the process of starting to integrate some of that structured data onto our graph so that we can be able to sort of correlate and tie them together.
Hiten: Awesome, last one from me in this topic. It's been fascinating to hear you guys exchange views on this. But the thing that strikes me, hearing you guys speak, is, it feels like we're in a little bit of an arms race where, like, the risks keep getting evolving and are more sophisticated, and the solutions kind of evolve and get more sophisticated. And often, the technology that's underpinning both of those is often the same. But David, paint me a picture, like we're in the 2030s, right, fast forward, you know, 10 years give or take. How does this all look then?
David: Oh, gosh! Great question! Well, I've been a big believer, and I've pushed this sort of concept for some time around augmented intelligence as opposed to sort of purely thinking about things as artificial intelligence, and I guess what I mean by that, which we're now seeing the real signs of in a very profound way, is the idea that these technologies become a very, very natural extension of the way that we work, the way that we make decisions, the way that we do business day to day. Much like in the last 10 years, we've gone from having, I don't know, e-commerce and Internet-based businesses to every business being digitally empowered. If you don't have an Internet strategy, you're probably not doing so well today. So it's become ubiquitous. And it's kind of odd to think. Oh, there are Internet businesses and non-internet businesses.
We're not going to have AI businesses and Non-AI businesses. Signal will probably have to drop the AI in our name because it will be entirely ubiquitous. Every business will be empowered and augmented by these technologies. Every business leader will have a natural extension, almost like a chief of staff in their phone, that is, on the hunt, looking for information, insights, and intelligence to feed into the decision-making process in a very natural and immediate fashion and a very dynamic fashion. But I use the word augmented because I don't necessarily believe that decision-makers are going to be replaced, or large swathes of the workforce are going to be replaced. I think it will be people who know how to use these technologies and know how to be supercharged and augmented, that will potentially replace the workforce of old. And so this comes back to that point around how incumbent it is upon large organizations to start tooling up, to start re-educating their workforce, to start ensuring that they have the right skills, to start getting the individuals within these teams to think differently.
You know, 10 years ago, or even five years ago, was a risks team or a comms team going to have a data scientist or a data analyst in their team? Probably not. You know. They probably would have thought you were crazy. And that's not part of their job role. But we need more data-savvy, data-native people to be working inside of these functions. And frankly, in 10 years time, everyone in that function is going to be a data expert. If they're not, they're probably not going to be able to keep up with the pace that you described before.
Hiten: I think what you described resonates strongly. That augmentation, the augmented intelligence. I think there's a pattern here, right? One of the other experts in this space that we had on the show, Bin Ren, who'd been in this space for a long time, also painted a similar pitch about this augmentation rather than wholesale replacement. So it's clearly a theme brewing there, and it feels like something that resonates when you describe it to me in any case. I'm going to keep us moving, going to shift gears slightly, and the final part of the show is kind of reflecting on some of your personal lessons learned across the journey, and we encourage leaders to reflect on what's one of the biggest challenges that you faced. What's something that you would call out so that listeners earlier on in their career, or even in the midst of their careers, would benefit. I don't know if there's something that springs to mind that you'd like to share on that.
David: Yeah. Gosh, I mean, I think probably the biggest learning is that every 18 months or so the business changes, you know, and the role is transformed, and you have to just keep learning and keep pivoting. You know the two words that spring to mind are resilience and adaptability. You need both of those things in spades if you're going to hold the course. You know the journey that we went on in our first iteration in the garage, you know, was one of a sort of discovery. You don't quite know why you're there. You're finding out the reason, the true vision, and the mission of the business. You're trying to figure out if the technology even works. And I remember running our first experiments before we had a user interface where we managed to sort of connect our models to the printer that we had, and we ran a newsfeed through these models, and we asked it to print out stories about Mergers and Acquisitions (M&A), and I remember coming out of the printer with all these headlines about M&A, and I thought, oh gosh!
Thankfully, the technology can work. So I've quit my job for a good reason. The next 18 months are about trying to find product market fit. And you know, the first five, 10 customers that we sold to were just a total random smattering of different clients from different industries and different use cases. And you start realizing, okay, we're going to have to zone in here on that ICP [Ideal Customer Profile], we're going to have to start finding true product market fit, true repeatability if we're going to get any form of scale. You hit a certain inflection point, and you realize you're going to need funding, raising that pre seed and that seed round. And you know, ensuring that you had the right metrics and the right evidence to suggest you could build something scalable and sustainable. You know, you get to 50-60 people and suddenly, you don't know everyone's name anymore. And it's hard. You have to start codifying the culture. And, you know, establishing the ways of working. Internationalization is probably the next big chapter. You know, when you hit 120 people and you start opening offices abroad, and you realize you can't quite retain the same culture by geography.
You have to let each office have its own sort of dynamic and its own sort of cultural independence to some degree. And on and on it goes, you know, so overlaid on top of that is all of the macro stuff going on, whether that's Brexit, or COVID, or US political unrest, or geopolitical conflict. And you know, inflation. Those things all have an impact on our business as it does to the clients that we're servicing. So you're sort of navigating through these pivots of your organization. And you're also having to be very adaptive to what's happening outside of your organization. And I think that's what makes it fun. You're constantly learning, and you're constantly evolving. But you also have to build a pretty strong muscle for, sort of, resilience, if you're going to hold the course.
Hiten: When you describe that for some reason, I've got this picture in my head of this boat trying to navigate and being bashed around the sea. But your boat's getting bigger as you scale through that period. As well as the seas become a little bit more navigable, but then all the stuff you've got to do around how you run the crew on the on the ship.
David: That's a very good analogy. And you know what's interesting is, in the early days, you have very little sort of shock in your, you know, shock flexibility in your system. You know, the bumps in the road, you rarely feel them. As you get bigger, you can, as you say, navigate those choppier waters. But at the same time, when you're smaller, you're very, very close to the problem space. You're very close to the edge of what you're building and the customers you're serving and so retaining that proximity is one of the key learnings I've had as well. There was a period of time, certainly, within Signal’s life, and I think many venture-backed businesses experienced this where sort of more was better, you know.
More people were better, bigger offices were better, more funding was better. And in a way, what happens is you start to be that sort of boundary between you and the problem space and the edge of your organization becomes larger and larger. And actually, one of my big realizations was lean is mean and it's good. And being closer to the problem space is better, and you can do more with less sometimes. And so that's certainly been an interesting learning that I think the whole industry has gone through post the pandemic, where everyone got a little bit carried away, as we know.
Hiten: Yeah. I mean, our take on that is being much more disciplined on what the sustainable growth rate of any organization is. I think what you just described when we reflect on it was kind of this arms race for growth and this absolute level and the immediacy. And you can end up making some short-term decisions that don't really benefit you from the longer term, right? And you know, you really want to be compounding like, actually, the world doesn't seem to kind of reflect on the long-term benefits of compounding versus being able to say, hey, I'm a rule of 50, or I'm growing at X, and I think that's kind of balancing that longer-term mindset is probably been the challenge.
David: Absolutely. Yeah. And look, building something sustainable and enduring is really hard, particularly when you're trying to grow something fast. I think, as a founder, or as a CEO, you have to be very cognizant of aligned or misaligned incentives. Your job is to, in a way, thread that needle and take everyone on the right journey but also be conscious of where those incentives may diverge. And you know a fund's job is to deploy capital and get the maximum return of that capital. And you know, particularly in venture, it sometimes means that they have a bifurcation of winners and losers. And they're okay to sort of take more risk. Whereas a founder, yeah, you want to obviously build the biggest company possible. But you've got other stakeholders that you're managing as well, and you're trying to make sure that, ultimately, you can build something that is truly sustainable and has longevity.
Hiten: Yeah, it's probably not asked in enough conversations around. Is this company growing too fast? Right, back to your point on the incentives, the alignment that element like. Actually, it's very rarely challenged on that side. Everyone's always too slow, and it's not moving.
David: Yeah, well, growth, I think, look, fast growth is always an incredible elixir to most problems. But I actually think maybe the question is, are we taking too much money? Do we need this amount of funding? Is this amount of funding going to embed the right set of behaviors and the right strategy? Or is it going to make us make lazy choices and easy choices? And I think when money was so cheap, businesses were able to make a lot of lazy and easy choices that didn't necessarily build great businesses. And when you have constraint, great innovation, and great strategy comes out of that. So I think it's less about top-line growth and more about have we raised too much cash? And it was very much a market, and I myself was signed up to it, where it was like the bigger the amount of money you'd raise, the higher the valuation you could espouse, you know, the better, and that isn't always necessarily true. You know what we're trying to build is great businesses. So let's look at revenue. Let's look at growth. Let's look at stickiness. Let's look at profitability. Let's look at your efficiency and your cost of acquisition. You know those are real metrics around business, not how much cash you've raised, or your sort of perceived valuation.
Hiten: Reminds me of that image in the playground where dudes are walking around talking about their one rep. max on the bench press. All of the elements of kind, of strength, and balance throughout the window. Final one, we always invite guests to throw and share the spotlight, so, calling out a company or an individual that you'd like listeners to be paying attention to.
David: Yeah, I think maybe if I could give two and cheat, and I'll sort of highlight two at either end of the spectrum. I'll give a shout-out to my chair, Archie Norman. He's been a phenomenal influence on me and my business. But, you know, I think, what's probably most recently, I mean, he's had the most incredible career over many decades building and turning around businesses, particularly ones that have been challenged or going through a business model transformation, and he's helped them achieve that turnaround with great success most recently and notably, obviously, what he's done with Marks and Spencer is phenomenal, you know. The business had dropped out of the FTSE 100. And it's absolutely rocking right now, and you know, one of the biggest things I've learned from Archie is just that coolness and calmness under pressure.
You know, we talked about risk, we talked about crisis. He is someone who never makes a mountain out of a molehill, and he always is thinking about how to approach problems and issues in the most pragmatic way. So that's one end of the spectrum, and then the other end of the spectrum, I'll give a shout-out to a good friend of mine called Finn McCabe. He's building a business called Thema AI, and what he's doing is almost equivalent to what Signal AI does. But with structured data sets, what he's doing is using AI and these large language model technologies to essentially understand bottom-up markets and industries and sectors as they emerge and then overlay, you know, very accurate and high-quality structured data on top of them to be able to help investors and consultants understand the markets as they emerge, as they're bubbling up. And yeah, I'm an advisor to his business and very excited about his technology and what he's doing. So he's maybe one to watch.
Hiten: Awesome. Well, Archie, thank you for getting David on the show. And, David, it's been fascinating, a really, really enjoyable conversation, and covered a lot of ground. So, thank you for taking the time and being so generous with your thoughts and reflections.
David: Pleasure. Thank you for having me. It was a great conversation.
Archie: Thanks, all.
This transcript has been edited for clarity purposes.
