The democratisation of AI tools is changing how companies are built. What once required teams across engineering, marketing, sales, and operations can now be coordinated by a single founder supervising a stack of AI agents. Tools such as n8n, Replit, and Lovable compress development cycles and reduce dependence on specialised labour. In practical terms, one person can now oversee work that previously required several junior or mid-level hires. To test how far this promise extends in practice, Vestbee spoke with Bek Ventures and Flashpoint, focusing on what they see on the ground as investors working with startups adapting to increasing automation.
Lean startup teams and solo entrepreneurs
A recent Economist article gives a case study of Solace, a generative AI startup helping users navigate grief and end-of-life logistics, which was founded and is run by a single person. The founder operates effectively alone - product development, marketing, operations, and early go-to-market are supported by AI agents via the AI-native incubator Audos. There is no traditional founding team and no early hiring plan, as AI is more of a collaborator rather than a tool in a traditional sense.
This idea of solo entrepreneurs building “solo unicorns” has gained some traction in the media, and recently, we have also observed some successful exits or acquisitions of companies with a very small headcount — Base44 is one of the examples, a startup acquired by Wix for $80 million with just eight employees. Still, investors’ view remains measured and far from reaching some overly optimistic conclusions.
As Márton Medveczky from Flashpoint puts it, in early stages, AI-enabled solo or ultra-lean teams can be highly effective. Strong engineers, supported by modern tooling, can reach product–market fit with far fewer people than before. This is especially true in B2B categories that never approach hyperscale. And the truth is that most startups don’t serve millions of users, and as they operate below that threshold, their technical and organisational demands for scaling are lower. In that sense, it cannot be argued that AI extends how far a small team, or even a single founder, can reasonably go.
However, the constraints often appear over time. As usage grows, security, infrastructure reliability, and system resilience begin to get more complicated and demanding. Flashpoint observes that many products assembled largely through AI tooling struggle under sustained load. Code that works for early users often fails when exposed to real-world complexity, adversarial behaviour, or regulatory scrutiny.
In the long term, scalability, security, and resilience still require human oversight. Most AI-assembled products don’t withstand true high-scale usage without dedicated engineering work. As companies grow, people inevitably become necessary to manage quality, infrastructure, and the reliability of core systems. There are teams and tools that operate with surprisingly small headcounts — for example, Fixie AI is one of the emerging companies doing well with a lean setup. However, these startups often don’t raise much capital because they don’t need it, which means they fall outside the typical investment stage we, at Flashpoint, focus on. The pattern exists, but it’s not yet the dominant model in venture.
AI outputs scale faster than their quality
AI can facilitate startups’ scaling and remove the bottleneck originating from the shortage of highly skilled labour - the overall judgment and strategic oversight can in turn, become more challenging. One person can now supervise multiple parallel workflows: product iteration, customer support, outbound sales, marketing experiments, and reporting, making attention to that valuable, scarce resource. With fewer people involved, more responsibility sits with the founder overseeing the system.
For sure, AI accelerates output — it can generate code, content, and workflows, but it does not guarantee sound architecture, security, or long-term maintainability. Forbes has even described AI agents as approaching the role of “strategic partners” capable of overseeing entire departments. Flashpoint’s experience is more restrained. AI excels at repetition and execution, but still requires close supervision when decisions interact or compound. Bek Ventures’ Principal Arda Kirkagac makes the point directly:
AI tools are accelerating product development, but the fundamental success factors driving rapid scaling haven’t changed. When building becomes trivial, everything else becomes critical: market insight, customer relationships, team building, strategic judgment.
There is also a structural dependency. As The Economist notes, the AI infrastructure enabling solo founders is largely controlled by major technology platforms. Just as cloud computing lowered barriers while concentrating value upstream, AI may leave startups dependent on systems they do not control. That dependency complicates defensibility and valuation.
Defensibility in an AI-first world
If AI collapses the cost of building, defensibility cannot rest on technical novelty. Flashpoint is explicit: “The technical ability to build a product is only part of the equation,” and increasingly, it is the least scarce part. Bek Ventures echoes this notion.
“AI tools are accelerating product development, but the fundamental success factors driving rapid scaling haven't changed. We're seeing founding teams move faster with smaller engineering resources, yet the companies that endure are still built on deep customer understanding, the talent they attract and strong execution,” says Kirkagac.
Other critical components of real-world defensibility that investors bring up are:
- high-quality engineering that goes beyond AI-generated surface-level code,
- deep understanding of user workflows,
- and a moat that has an embedded positioning in customer processes.
So in turn, what matters is no longer whether one founder can build fast - we already know they can - but the real issue is what comes after and whether what they build can prove dependable and hard to replace before others, armed with the same tools, replicate it.
What solo unicorns mean for the ecosystem
Despite speculation about fully automated companies, Flashpoint does not expect AI agents to replace early human teams outright. Instead, low-level, repetitive tasks will become (if they are not already) fully automated. High-value creative work remains difficult for AI — strategy, product vision, and contextual decision-making do not scale cleanly through agents. And because AI lowers the cost of early progress, expectations rise. Flashpoint observes that “the bar for metrics has risen.” What once counted as traction is now baseline.
At an ecosystem level, we can observe that building and scaling a venture has become more democratized; however, running a very large organisation with only a tiny team is hard to imagine. Ecosystems compound value through talent circulation, institutions, and repeatable company-building. Ultra-lean companies contribute less to these dynamics.
