Yariv Adan, General Partner, ellipsis venture
There's never been a better time to be an AI developer. If you combine technical chops with a sense of product design and a keen eye for automation, you're likely to build a very useful app during a hackathon weekend. So, is it time to stop VCs? Conventional wisdom says that if you can find a market gap, deliver real value, and ship quickly, you have a recipe for a business-based startup. You've probably watched many of your peers do just that. But before you join the multi-billion dollar unicorn hunt, you should ask yourself: would you be better off herding donkeys?
and beginnings change. Not incrementally, but fundamentally. Over the past year, we've met team after team doing everything right: moving fast, building useful products, targeting real customer pain, delivering real value. And yet, we passed on many of them. Not because the parties were weak, but because the channels that would have protected their importance have been completely destroyed.
The most basic rule of business has not changed: a company needs diversification and secure channels to continue to achieve high margin at scale. But what matters as a secure channel has changed a lot, when the bar rises to the highest level. If your business lacks a real channel, whether it's proprietary data or unique information that can withstand an army of highly skilled AI agents, it will inevitably face disruption within the asset execution space.
Two years ago, we invented this name Commoditized Magic to explain the future we saw the drawing of AI. Technologies and products become truly magical, unlocking capabilities that were previously impossible yet almost entirely consumed by frontier models. We are always optimistic about the “magic” part: it presents a huge economic opportunity by unlocking previously inaccessible value. But the risks of commodity sales are real and disruptive, making all areas uninvestable.
In this piece, we want to unravel that dynamic of the commodity: why the unicorn is so difficult to hunt in the current environment. But we also want to suggest that a new creature, or rather, a very familiar one, will appear: herds of donkeys.
Goods from all directions
AI consumes software and services, but at the same time, the economics of the unit of value creation is changing dramatically. The cost, technology, time, and resources required to bring a product to market are decreasing. That changes everything, and sales are running in all directions.
User as builder. There is a new category of apps that replace previously purchased software: the ephemeral app. Whether it's a simple prompt that creates an artifact, a Claude code session, or a combination of skills, tools, and plugins users can now build any application they can imagine. Any experienced developer knows that building even the most complex single-user, one-time module is trivial; Traditional complexity and expertise come into play only if they make it modular, general, scalable, and maintainable. A single user builder is a formidable competitor to every SaaS company when it comes to building the application they need at a given time. This extends to groups as well, and to organizational memory, beyond that.
An explosion of rivals. As coding agents evolve and reach the level of professional human developers at much lower costs and management complexity the barrier to entry to becoming a SaaS company drops significantly, resulting in orders of magnitude of competitors. The result is congestion at all levels, and we're already seeing it in our app. Every use case now has many startups attacking it, each starting in a small area of the ocean where it has a negative opportunity, hoping to expand and win the market. But when they raised their heads, they saw the heads of the sea around them, with no clear difference. These companies can bring real value, some may be profitable but they don't make sense as venture-backed businesses.
Venture and startups have always been a numbers game of hits and errors. But when the scales change by orders of magnitude, with so many companies, lone inventors, and small teams all empowered by the same tools, the old rules break down. You end up with more misses than hits, until the VC model itself stops working.
“It's All About Distribution” Or Is It?
The argument we often hear is that in a world where software is a commodity, it's all about distribution: move fast, capture those first customers, and you win. Unfortunately, commoditization and AI are rewriting the rules of go-to-market and distribution once again.
First, there is the problem of congestion. If you can move fast, make an MVP fast, then sign a pilot, all in four weeks with two people, and so do many of your competitors.
Second, AI not only unlocks ephemeral, human-made applications, but integrating traditional software has also become much easier, faster, and cheaper. Traditional SaaS products come standard and require complex, expensive integration projects, a large source of adhesion and initial profit. In the new world, where this integration can be done automatically or renewed on the fly, those channels quickly disappear. As lock-in effects weaken and the customer no longer needs to worry so much about future support and compatibility, they can focus on what they need now, and who does it best, especially in highly competitive and competitive markets.
As a result, we expect AI software agents to emerge to replace the old human-led methods. These agents can bid and evaluate in real time the skills needed, threatening to render the product, distribution, and first mover profits irrelevant. The economics are clear: when switching costs approach zero, reliability follows.
Finally, Big Tech is moving up the stack and moving beyond verticality. Imagine how the providers of the boundary model and the owners of the platform, think of email, chat, and documents in business, or mobile, search, and social communication for consumers, can now create specific use cases, faster and better than before. Google is adding AI capabilities directly to Workspace, Microsoft is embedding Copilot throughout Office, Apple is integrating intelligence into iOS. These giants are moving into what was once a startup space, leveraging the benefits of distribution for startups that will never be the same. The ability to develop at a very high speed applies to Big Tech as it starts with two people, and Big Tech starts with a billion users.
This is the new reality in the software and services market, as useful intelligence becomes a commodity.
Donkeys, Not Unicorns
Is this the end of business, isn't there a way forward for small, strong teams that can bring immediate value to underserved markets? Far from it.
There is clearly a huge opportunity for new unicorns, with just a higher bar. That is the opportunity we are focused on as a VC. But we also believe that the great power and speed of AI has opened up another way for entrepreneurs, which does not require any venture capital.
What if, instead of chasing that one elusive unicorn, you used agents and low development costs to automate and scale the creation of value-generating businesses? Can a single founder build a herd of low-income donkeys at scale?
Imagine what that looks like in practice. Automate ideation and market research to generate, prioritize, and prune the pipeline of ideas. Conducts user research and interviews, customer outreach, hypothesis generation, prototyping, testing, and analysis. You open these businesses, run them in parallel, kill the losers, double-team the winners, and adapt them as needed.
Imagine a founder running fifteen small businesses at the same time, each serving a small niche targeting an underserved market segment he has access to: one automates compliance reporting for small European fintech firms, another produces custom training materials for logistics companies, a third manages invoicing workflows for independent consultants. Almost even with the focus of the country. None of this is a multi-billion dollar market. None of them will make it to the TechCrunch article. But each generates a steady, steady income, and together they add up to something meaningful. The founder does not manage fifteen teams; AI agents handle build, iteration, customer support. The founder's job is to manage the portfolio: which donkeys to feed, which to retire, which to enter next.
This is the opposite of the business model. Instead of concentrating the risk on one big bet, spread it over many smaller ones. Instead of needing a 100x return on a single company, you build a portfolio where the combined effect is what matters. The calculations are different, the risk profile is different, and most importantly, it does not require external capital, which means that the founder retains full ownership and control.
We recommend this approach to the teams we meet who are doing great work but are working in areas where the moat is not deep enough to have a business scale result. Often very small and efficient, these groups are better suited to bootstrap than scale up. The way of the donkey is not a consolation prize. For many founders, it may be a smart play.
This isn't a business-class game, and that's exactly the point. It's a new way for entrepreneurs who are willing to trade the dream of one big result for a smaller, sustainable portfolio, and use AI to make that portfolio manageable to a degree that wasn't possible before.
We believe there is a real opportunity here, and we have started testing the tools to make it work. Stay tuned.
