We spent some time last week reading Anthropic's new Founder's Playbook: Building an AI-Native Startup. It's one of the more practical AI documents we've seen recently. Less about AI, more about what founders should actually do differently.
The obvious takeaway is that AI is changing how tech companies are built. The more interesting question is what that means for entrepreneurs, investors and the broader HealthTech ecosystem.
These are our five takeaways.
1. The founder's job is becoming orchestration
One line from the deck stayed with us: "The lean 10-person unicorn has gone from scrappy underdog story to deliberate plan of action."
Most people will read that as a productivity story. We think it's a role-change story. For years, founders were defined by what they could personally do: write code, close customers, hire teams, run operations. Increasingly, they are defined by what they can direct. Ideas go in. Agents, software, and a relatively small human team execute. The implication is that headcount is rapidly losing its value as a signal.
For more than a decade, investors treated team size as a proxy for momentum. A company with five employees looked fragile. A company with fifty looked real. That shortcut is becoming less useful by the month. Today, a two-person team with a thoughtful agent stack can achieve what previously required twenty people.
What interests us is not the productivity gain itself, it's the type of company this suddenly makes viable. A founder building scheduling software for outpatient clinics, workflow software for diagnostics providers, or compliance software for pharmacies can now reach meaningful revenue with dramatically less capital. In many cases, profitability becomes achievable long before venture-scale growth.
Historically, these businesses often sat in an awkward financing gap. Too small for venture. Too capital-intensive for bootstrapping.
AI will substantially increase the number of companies that fit exactly this profile.
The irony is that the more successful AI becomes at reducing operating costs, the less natural the venture model becomes for a large subset of software companies.
2. Geography matters less than it used to
The deck repeatedly recommends using AI for tasks that founders previously solved by finding an expert.
Legal questions. Product decisions. GTM planning. Research. Infrastructure.
For years, access to expertise was one of the hidden advantages of startup hubs. The answer to many early-stage problems was simply knowing the right person. That advantage is shrinking. Many aspects still matter - distribution, brand, networks, procurement etc. But on the build side, the gap is compressing.
Anthropic highlights examples such as Carta Healthcare using Claude to support clinical abstraction workflows. Whether or not every case study ages perfectly, the broader trend is difficult to ignore: AI is making specialist knowledge dramatically more accessible.
This is particularly relevant in Central Europe. The region has had strong engineering talent for years. It has also produced a steady stream of healthcare founders with deep domain expertise. What it often lacked was the surrounding ecosystem that existed in places like London or Berlin. The build-side disadvantage is becoming less significant. A clinician-engineer building software in Prague, Dresden, or Kraków is increasingly competing on the quality of the problem they solve rather than on their proximity to a major startup hub.
The distribution challenge remains, bnut for the first time in a long time, the starting line feels more level.
3. AI reduces friction and increases the cost of poor judgment
One of the strongest sections of the deck is about failure modes. Anthropic describes confirmation bias as something that now comes equipped with a research engine. AI makes it easier than ever to find supporting evidence for an idea you already want to believe. The output looks rigorous, the citations look convincing, the market maps are comprehensive. And yet the underlying thesis may still be wrong.
We see this increasingly in HealthTech. The decks are often better than they were two years ago. The market sizing is cleaner. The competitor analysis is more exhaustive. The regulatory section is better researched. But many still avoid the questions that actually determine whether the company works. How does the product fit into clinical workflow? Who changes behavior? Who pays? What happens at renewal? Where does reimbursement help and where does it hurt?
Those questions are harder to answer because they require experience rather than synthesis. AI has made polish abundant. Judgment remains scarce. And in regulated sectors, judgment matters far more than presentation quality.
For investors, that makes domain expertise increasingly important. Not because AI is replacing diligence, but because it makes superficial diligence easier to fake.
4. Validation is becoming more valuable than velocity
The most useful parts of the playbook focus on structured skepticism: Ask the model to argue against your idea. Build the strongest possible case for competitors. Look for evidence that disproves your thesis rather than evidence that supports it.
Most founders are naturally optimistic. Most investors are naturally attracted to optimistic founders.
The challenge is that AI amplifies whatever direction you point it. If you ask it why your company will succeed, it will generate reasons. If you ask it why your company will fail, it will generate reasons.
The quality of the outcome depends heavily on the quality of the question. This is one reason we remain particularly interested in operator-led founders. A practicing clinician building software for clinicians. A pharmacist building pharmacy software. A hospital operator building scheduling infrastructure. They have a natural advantage during validation because they already understand the environment in which the product must survive. The same logic applies to capital.
The most valuable investor in these businesses is often not the one with the largest network. It's the one capable of challenging assumptions at the level of the domain itself. In healthcare, pushback is often more valuable than introductions.
5. The work that compounds is the work AI doesn't eliminate
Another theme runs through the playbook: AI dramatically reduces the cost of building, but it does not eliminate the need for discipline.
Anthropic spends considerable time discussing specifications, architecture, documentation, and project memory. Without clear principles, teams end up rebuilding context every session. Without architectural discipline, products accumulate technical debt faster than they can create value. Without clear decision-making processes, founders become bottlenecks.
The same principle extends beyond software. At the MVP stage, founder involvement in every decision can be a strength. At scale, it becomes a constraint.
The deck suggests mapping every workflow, approval, and decision that flows through the founder and identifying what would stop if they disappeared for a week. It's a useful exercise because the answer is usually uncomfortable. The companies that benefit most from this discipline are often the least glamorous ones - clinical workflow software, pharmacy infrastructure, compliance systems etc. Administrative tools embedded deeply into day-to-day operations rarely generate headlines, but they often generate retention. And retention remains one of the few advantages AI cannot manufacture.
Outlook
Our biggest takeaway is not that AI will create more unicorns - it's that AI will make it possible to build a lot more good businesses. Particularly in sectors like healthcare, where complexity, regulation, and fragmented markets have historically made venture economics difficult. A clinician building workflow software for a narrow specialty practice may never become a €1B company. Ten years ago, that often meant the company wasn't financeable at all.
Today, the same business can be built with a fraction of the capital, reach profitability faster, and generate attractive long-term returns.
We have spent decades building capital structures optimized for the rare outlier. AI will dramatically increase the number of companies that don't fit that model - and are still excellent businesses.
