The #1 AI for Founders & Investors: Pixie Beats 95% Startup Failure Rate & Lifts Global Economies

Arham: Welcome to the AI and Beyond for Business podcast. Today we’re going to be tackling a very tough question: Why do so many venture-backed startups fail? More importantly, we’re going to explore a groundbreaking new solution that’s aiming to change that statistic for the better. I’m thrilled to be joined by two very special guests who’ve been instrumental in this innovation. First, we have Richard, the founder of HeyPixie, an AI-driven platform that’s taking a unique approach to startup success. And joining him is Keith from Zazmic, representing the engineering powerhouse that provided the resources to build Pixie. Richard, Keith, welcome to the show.

Richard: Thanks for having me. It’s great to be here.

Arham: So Richard, I want to start with you. Let’s get straight to the heart of the matter. The vast majority of venture-backed startups end up failing. The number I saw a few years ago that was popularly going around was about 80%, 83%. From your vantage point, what’s the single biggest reason that this failure rate was so high in the first place and has only continued to rise?

Richard: The tragedy of the situation is that that statistic hasn’t really changed. It’s a very narrow margin, whether it’s 15% or 10%. To some extent, depending on which part of the world you’re in, that will have a big impact on the success rates. The figures show well north of four out of five don’t make it. And we’re talking about companies that had the intent to scale, not just all new companies set up. That’s a very depressing number if you consider the enthusiasm, the energy, the innovation, and the capital that’s gone into getting these businesses to the starting line in the first place.

Arham: Right. With this massive problem as the context for our conversation, can you introduce Pixie? In the simplest way, how is your platform designed to be the antidote to this problem you’ve described, and what makes it so unique?

Richard: The problem’s not hard to spot—it’s an enormous problem. The issue has been the way we tackle it. We have fantastic, brilliant people working in venture capital firms, ecosystems, and incubators who help startups be more successful, and the impact they have is dramatic—sometimes doubling or tripling outcomes. But their bandwidth is very narrow. We’re talking single-digit capacity; at maximum, only 5% of UK startups can get onto an incubator program, and about half that get VC funding. Our view was, that’s fantastic, but how can we take those principles—the wisdom, the insight, the ability to nurture—and bring that to a broader church? Essentially, democratizing the opportunity for individual companies to be more successful. To scale dramatically, you need a system. So we’ve built a system to manage scaling on a systematic basis. Like any large ecosystem, whether natural or commercial, if it’s large enough, there will be patterns. We set about decoding those patterns in the ecosystem. In a Darwinian sense, what separates the winners from those who fail? If we can understand that delta, we can work on it. If we can narrow the delta, more will scale up. It’s as simple as that.

Arham: Richard, I’d like you to dive a bit into what Pixie is doing from a functional standpoint. How do both users—investors and startups—come together to gain some benefit from this solution?

Richard: The lovely part of this is that both sides win. The more startups that successfully scale up, the more investors will generate returns. This is part of the bridge Pixie is forming between companies looking to be invested in and scale, and investors looking for great companies to fund. There is an alignment of interest between the investor and founder communities. The process we use to bring them both along aims for an outcome that’s in everyone’s interest: higher success rates.

To make this work, we went back to fundamentals. There are immutable laws in the startup universe—patterns we can decode, identify, and empirically validate. We can academically ground the behaviors and traits common to successful companies. Once we isolated them, we had about 100 benchmarks we were happy with, and we’ll aim for 250 by the time we launch. This gives us a whole raft of different ways to reference individual companies across criteria like financial management, market presence, hiring policy, location, and team incentives. This all feeds into what we call the Scale-Up DNA. We’ve decoupled what it is to be a success.

We then take each company on Pixie and ask, “How do you measure up against these benchmarks?” Based on that, we figure out which gaps we can help you close now, which can be part of the future, and how that plays into your growth journey. This melting pot—this data lake—is something Keith often pales at. We are looking at a quarter of a billion potential data points at the intersection between these benchmarks and our data sources. We use Vertex for APIs into commercial databases like Crunchbase or Dealroom, pulling open-source data from sources like Companies House or the Office for National Statistics in the UK.

When we intersect, we have facts—a solid starting point for the AI to work from in doing the diagnostics, reasoning, and inference. The data we’ve gathered is all hindsight—what happened. In the middle, there’s Insight, where we decode that data with an algorithm and intelligence. Then we bring that into a forward-facing version of Pixie, which tells you, as a customer: Given your circumstances and objectives—these are the five things right now you need to be focused on. Over a typical scale-up journey of four to seven years, you’ll probably make about 5,000 such considerations. The math is straightforward: if you make each one of those 1% better—a tiny incremental improvement each time—your chances of being successful have just gone up 50%. It is that simple. Small gains, huge outcomes. That’s our approach, and it’s completely different from how startups are currently supported.

Arham: I appreciate that response, Richard. We’ll come back to a more in-depth conversation about the solution and the technology. I want to zoom out and bring Keith in to ask about his experience of that first conversation with Richard. How did the idea evolve over time, and what was your initial reaction?

Keith: We had such an interesting start to our discussions, Richard. Our first one-hour introductory call began with a company introduction, but we spent the next 45 minutes doing live demonstrations of AI use cases and the POCs we’ve built in-house. This accelerated the collaboration, moving from the what if to we can deliver flexibility with AI solutions, aligning perfectly with the Pixie goal. That accelerated the refinement of the delivery strategy, SOW creation, and mapping out the roadmap to kick off the iterative delivery processes.

Richard: Technical agility has been a fundamental part of our philosophy. The things that truly interest me are the things you can’t do—because you absolutely could not have done Pixie 18 months ago. Talking about what is possible helps us develop what will become possible. I don’t exaggerate: the only hardware the company owns is the laptop we’re using right now.

We’re not raising vast amounts of capital to buy huge servers or spending millions training language models. We’re much better off working with companies like Zazmic and Google who have access to those resources. The combination is tremendous. That’s what I love about working with Keith: it’s not “what are we going to do,” it’s “how are we going to do it?” I’ll say, “I want to do this,” and Keith will say, “Well, you can’t do that.” I realize that, but if we could, how would we do it? We can now deliver Pixie at an economic cost; it’s borderless and accessible to all. We just bring solutions. I love that we can offer Pixie not only in obvious locations—it’s almost slightly unobvious to launch in London, the world’s number two-ranked startup ecosystem. If we can prove it here, we can prove it anywhere. The ability for us to take all the great things we’ve learned and deploy that into places with enthusiasm, courage, and intent is really exciting. The CEO of Loveable recently spoke about how the technical hurdle for an AI-native business has been reduced so dramatically that he can see a 50x increase in the number of AI-driven businesses. That’s fantastic! Because the technology stands to one side.

There are still going to be challenges with scaling a business that have absolutely nothing to do with having the right engineers or programming capabilities. Founders still have to scale that business and deal with challenges that are outside the engineering team. This is where we enjoy a certain future-proofing with our business model. We’re not anchored in technology, hardware, or solutions that will be easily overtaken.

Keith: I think the time is perfect for Pixie and your vision, Richard. We have a perfect synergy of economic infrastructure that can be molded to deliver the vision. We also have this inherent VC or startup mentality that spans back to the 1970s. We see innovation globally hitting a rapid ramp. If the traditional evolution of startups isn’t addressed with an innovative approach, there will be a mismatch between where innovation is running and the inherent growth patterns of these companies. It seems like it’s the absolute perfect time for Pixie.

Richard: We could call it a competitive white space. On one side, you have the software technology, usually SaaS platforms in startups; on another, you have the smart people with long experience nurturing companies to enormous valuations. And then you have the extraordinarily intelligent, expensive business intelligence platforms that are beyond the reach of most startups. But in the middle is this huge vacancy, and we’re driving straight at it. We’re not competing head-on with the SaaS platforms, the VCs, or the big enterprise intelligence platforms. What we’re doing is fixing the middle bit, where founders right now are not well-served. We are bringing a valuable service to them at a price they can afford, on an accessible platform. We can absolutely improve their potential to scale their business.

Keith: This has a positive impact in terms of the evolution from a founder’s perspective and a positive impact from a VC’s perspective. Then we also have the hyperscalers—Google, Microsoft, AWS—building infrastructure and delivering IaaS and PaaS services. They all have specific startup programs. The domino effect doesn’t just stem from a founder’s evolution; it directly impacts VCs and hyperscalers. The knock-on effect goes right up through the chain of growth.

Richard: On top of that, there’s a network effect. The more successful we are, the more we understand what success looks like, and it becomes a virtuous circle. One interesting question I’m often challenged with is the tipping point where Pixie has developed enough knowledge to provide its own intelligence. If we think of having 5,000 companies onboarded onto Pixie next year, and we feed that into the data pool, you get into huge data intersections, nudging up against the quadrillion range. At some point, there will be enough capability, mapped with the vastly accelerating abilities of AI, to then synthesize genuinely unique advice based on a knowledge base we almost entirely control. 

Arham: You talked about why startups fail—it’s not necessarily the idea, but the missing pieces. There is also the second side of the coin: the investors. A great startup can fail if it’s not connected to the appropriate investor. Could you talk about how Pixie is a solution not just for startups, but also for investors?

Richard: There is an absolute coexistence and co-alignment of interests here. A typical mid-size VC will probably invest in about 1% of the people that come across their threshold during the year. From first contact to cutting a check, there’s a 99% inefficiency in that funnel. We can bring a much smarter alignment between people seeking capital and people providing capital.

This is where the Pixie Scale-Up Score is key. We can bring to our Scale-Up Score a similar degree of confidence and intellectual integrity that you’d get in a credit score. It will become a benchmark score that will be adopted by the industry as the best reference for a startup’s scale-up potential. We can go to a VC and say: “We understand your mandate; you only want to invest in B Corp companies. We’ll agree that you’ll take a first meeting with anyone who has a Scale-Up Score over 600.” We can confidently say we can narrow that funnel down by two-thirds, leaving them with the 33% they should be talking to. That’s a massive cost saving in terms of deal efficiency.

Once companies are in a portfolio, there’s a lot of noise. Some companies are very good at making noise and are the ‘alphas’ who will likely get second checks, but they’re not always the winners. Other, quieter companies, more focused on the business, can be overlooked. Harry Stebbings beautifully describes this as the ‘messy middle’. Inevitably, the companies he thought would be the money-makers are not, and the ones he was going to ‘get around to’ are the ones that hit the lights out. We will bring a system to that mindset, taking the guesswork out of it. Pixie will shine a light in dark corners and reveal overlooked opportunities. The impact of a general scale-up rate increase is felt even more when Pixie is put within a managed portfolio environment.

Arham: What you’ve pointed out is quite interesting. Even with all the incubators, VCs, and TV shows like Dragons’ Den and Shark Tank, it’s a highly unstructured market. It almost feels like success is left up to chance and luck, rather than the foundational strength of a startup to scale up.

Richard: Interestingly, for Shark Tank, you have a 600% better chance of being successful if you’re invested as part of that program. But they only invest in maybe 30 or 40 companies a year. We have 400,000 startups trying to scale in the UK alone, and 7.5 million in the world. Shark Tank is not going to make a dent in this.

Keith: A drop in the ocean.

Arham: And out of the 30 or 40 that get funding on air, half of them drop out. Many might also just be there for the marketing, so there’s a mix-up of the actual purpose.

Richard: Yes, I hear the due diligence process can be troublesome occasionally.

Arham: Absolutely. It’s great that something like Pixie is trying to address this major global issue. I want to understand what led you to take the decision to go after this problem.

Richard: Professionally, I’ve been involved in data almost since the beginning. Data-driven marketing has been a constant theme; the idea of using data to reveal asymmetrical opportunities is where I was brought up. As an entrepreneur, I’ve been through the school of hard knocks, having both great wins and painful losses. The mistakes I made might have been avoided if I had someone looking over my shoulder.

I got involved in financial modeling for a new early-stage VC firm, which exposed me to the fundamental mathematics of how it operates. The expectation is that only a tiny number of companies will successfully produce returns. From an entrepreneur’s point of view, I thought, that doesn’t sound very democratic. What could be done mathematically to take the sting out of that curve? Then, I was handed some research that showed a close alignment between startup behaviors over time and their propensity to be successful. I thought, there’s an idea here. If the principle of looking for patterns of success—the Scale-Up DNA—could be decoded and applied at scale to every startup on the planet, bringing small increases in performance at each step of their journey, that has the potential for real impact.

To put it in context:

• If we increase the global startup scale-up rate by just 0.1%, that would unlock about 7,500 more successful scale-ups, create half a million jobs, and generate enough income tax to fund the building of 1,000 schools in non-G20 countries.
• If we aim to double the rate from the current global average of 4.6% to London’s rate of about 9%, the tax generated by those successful businesses would wipe out illiteracy up to the age of eight in five years.

The impact on society goes far beyond the interests of the founders or the capital investors. There is enough amplitude in this that if Pixie is successful, it leaves a legacy of benefit for all corners of the planet. The scale of the unrealized potential is staggering. If you take all the startups in the world and say only 4.6% will be successful, the value of those that are not successful is equivalent to the most valuable asset class in the world: real estate. Can you imagine an asset class of that scale operating on a 4.5% success rate? It’s insanity. The time is now, and somebody needs to do something about this.

Keith: The current market outlook, combined with the decentralization of business launching that the Loveable CEO spoke about, is significant. The large hyperscalers are always bringing innovative products to market, but what’s happening with startups is that he’s estimating a 50x increase in the sheer volume of new businesses. The brightest minds and applications are in the minds of the 90% of the population. We are at a complete convergence between the acceleration of innovation and economic, pay-as-you-go cost models for services like AI that you can customize. I feel the number of startups in the UK alone will probably double or quadruple over the next 5 to 10 years.

Arham: With the number of startups we’re going to see over the next few years, particularly with AI in the picture, Richard, how do you see Pixie catching up to that increased rate in the next five years?

Richard: Part of me wants to say that the rate of success is so poor that the last thing we need is more startups. But the other side says, if we could increase the quality of the starting points—the smartest ideas don’t necessarily come from the people who can write the code. We are reducing the barriers for smart ideas to come to market. That area becomes a lot less terrifying because of the reduced upfront commitments needed to get built. Look at the way we’ve built our product with Zazmic: it’s been painless and fluid. The size of our technical team is tiny because we work with people like Google and Zazmic. Our interface with them speaks to the agility we want to bring to other startups. Our burn rate is lean. We’re spending the money in a smart way, exactly when we need it. The balance between go-to-market, product development, and resource building are constantly evolving metrics. Ironically, we are Customer Number One for Pixie, so we have to believe that Pixie can help us be successful in the same way we’ll make our customers successful.

Arham: On the point of technology, Google Cloud has allowed you to be very lean and utilize Zazmic’s engineers to build the platform. For both of you, what was the experience like applying the Pixie idea onto Google Cloud’s technology, and what made you lean into Google Cloud as the right place to be building?

Richard: In Google’s Shoreditch studio, one of the smartest people I’ve ever been in the presence of was talking about the upcoming release of Vertex. Everyone in the room was a coder apart from me. I literally got a piece of paper and sketched Pixie, asking, “Can we do that?” He said, “Not now.” I asked, “But could we do that?” He told me to go speak to the Google startup team. They took us on board, which was a flattering experience, and put us in touch with Zazmic. That’s where Keith came in and made sense of all my rambling.

Keith: From there, it was about building a solid, open, and iterative partnership. Putting proof and execution into the ability to deliver with the solutions and services really reinforced and accelerated the delivery methodology. We had consistent daily communication, tackling phases with weekly sprints, demonstrating previous sprints and forecasting the next. It was a very granular, iterative approach. The beauty of Richard and the Pixie team was that they were incredibly well-prepared in terms of expectations. Our team took that vision and broke it down into a phased, bite-sized approach, working collaboratively through the delivery. It was an absolutely amazing experience.

Richard: We’re only just getting started. Firing up the jets.

Arham: Now that Pixie is built out, what would success look like to you for the next 5 to 10 years?

Keith: I would like it to be an industry standard.

Richard: If Pixie becomes the immutable center of gravity—the reference point that courageous founders go to make the most of their ideas—it will be a ubiquitous resource. That’s a beautiful way of framing it.

Keith: I always refer back to the hyperscalers building out their cloud platforms; their vision was that it should become an industry standard. If they hadn’t achieved that, the global economic patterns we see now wouldn’t exist. They created economic pay-as-you-go models, global deployment, and delivery, and all those evolutions made it an industry standard. The historical trend of VCs and startups has seen dramatic changes, but has there been a specific flip of the coin where it becomes an industry standard? Not yet. Look how fast innovation has taken us by storm. I see Pixie becoming an industry standard. Let’s legislate for it.

Richard: Even spiritually, being a founder is a lonely existence. I call it the loneliness of the long-distance founder. What Pixie can bring to this ecosystem is the sense that you’re never completely alone. It will be a place to congregate for the obvious things like capital, data, and performance optimization, but also for humans to interact on a common subject. The level of connectedness across founder networks outside your own country is a very strong determinant of success. I like the idea of building something with a community dimension, a community with a purpose, so founders don’t have to feel alone.

Arham: Fantastic. I love the idea of making this a timeless piece of technology. For the many startup founders in the thick of this industry right now who may be feeling lonely, what piece of advice would you give them in navigating the space?

Richard: Firstly, I’d love for them to reach out to me at richard@heypixie.ai. I’m genuinely keen to build an initial program of test pilots—people I can develop Pixie with to make it the best version of itself. I’m genuinely looking for people from all corners of the planet to talk to. If you’re watching this now and would like to be involved, visit the website heypixie.ai or find Richard McCandless on LinkedIn.

Arham: We’ll attach that to the show notes so everyone can access it. And for those same startup founders, what’s one piece of advice for navigating this industry right now?

Richard: Take the guesswork out of it. Make your journey as objectively informed as possible. This is not a substitute for your courage or innovation, but there is a way to narrow the landing slope, to get the plane to land in a more predictable fashion. This is us helping you grind it out until you reach a point where you’ve been successful. I am absolutely sure we can improve the chance of that happening.

Keith: Love that!

Arham: Awesome. I think that’s a wrap. Thank you so much, Richard. Thank you, Keith, for joining us today. And thank you to everyone who’s listening. We’ll attach Richard and Keith’s contact information into the show notes. I hope you had a great time listening. Until next time, this is the AI and Beyond for Business podcast. Bye.