The floor is lava. The ceiling is breaking. And somewhere in between, AI startups are racing to define what success actually means.
In a market where the pace of innovation is matched only by the speed of irrelevance, even doubling year-over-year growth might not guarantee a second meeting. Founders face an unforgiving calculus: grow faster, differentiate deeper, and embed smarter — or get swallowed by the next shiny thing. The AI gold rush has evolved beyond model performance or technical wizardry. Now, it’s about ownership of workflows, data defensibility, delightful user experiences, and value creation that’s both obvious and impossible to replicate.
At our April AMA with Madrona portfolio company leaders, we explored this shifting definition of success in the AI era. What emerged was a blueprint for how exceptional companies are navigating market volatility, rising above the noise, and building with urgency — all while raising the ceiling and lowering the floor of what their products can do.
From services-as-software and private data moats to market consolidation and IPO signals, here are the most urgent takeaways founders should know now.
Success in AI: Raise the Ceiling, Lower the Floor
In AI, it’s no longer enough to serve just the technical elite. The most impactful products are the ones that make the best users better — while also unlocking entirely new categories of users who never would have accessed the functionality before.
This is how companies like Cursor are pulling off unheard-of growth trajectories: helping expert developers ship faster and enabling non-technical users to build software for the first time. It’s the magic of expanding TAM at both ends of the spectrum. If your product makes power users feel like superheroes and first-timers feel like insiders, you’ve cracked the code.
This “dual wedge” strategy — elevate the top while empowering the bottom — isn’t just clever. It’s how category leaders are born. It also speaks to why even “consumerized SaaS” in enterprise environments is seeing such velocity: ease of use is as important as power.
The Floor is Lava: Good Enough Doesn’t Cut It Anymore
Even amid uncertain macro conditions, capital is available for breakout companies — but the bar is higher than ever. Founders of companies in single-digit million revenues might expect that 3–4x growth would earn them a second meeting, but the reality is more complex.
Capital is still out there — but it’s gravitating toward the top decile. Growth alone isn’t enough. Investors are looking for:
- High growth velocity
- Evidence of defensibility
- Clear ROI and user love
- Efficiency paired with scale
They’re asking: Are you defensible? Are you indispensable? Are you irreplaceable?
In a market flooded with tools, most of which look similar on the surface, only a few stand out. Disruption is increasingly bottom-up. Tools that start as $20/month developer favorites are gaining enterprise traction before traditional players can react. And unlike SaaS 1.0, these new entrants often skip procurement entirely — growing through bottom-up adoption and surfacing only after they’ve embedded deep within teams who become raving champions.
For founders, this means two things: Your product has to be incredibly good from day one, and it has to earn the right to stick around. Switching costs are low, and loyalty is even lower. What matters most is velocity — of adoption, of iteration, of value delivery.
Own the Data, Own the Outcome
In AI, infrastructure is increasingly commoditized. Foundation models are powerful — but also interchangeable. As the pace of model innovation gets increasingly frenzied, durable companies find long-term advantage not by model performance but by how deeply they manage customer data and how well they deliver a game-changing workflow and user experience, transforming how the user works.
What truly separates companies that endure from those that fade isn’t better models — it’s better data. More specifically: it’s exclusive access to proprietary or private customer data, embedded inside real workflows, that unlocks defensibility and long-term value.
This is where moats are being built in 2025.
Breakout companies aren’t winning because they’ve secured the best infrastructure. They’re winning because they sit closest to the moment of value creation — where decisions are made, documents are created, code is written, and insights are formed. That proximity gives them access to high-signal, workflow-level data that’s unique, compounding, and impossible to scrape or replicate.
When your product collects that kind of data:
- You unlock smarter, more personalized model outputs.
- You build feedback loops that improve performance every day.
- And you raise the switching costs dramatically — because the product isn’t just useful, it’s trained on the customer.
Whether you’re building a vertical or horizontal play, this is non-negotiable. In verticals, embedding with proprietary data accelerates adoption and ROI. In horizontal platforms, that data is what gives you defensibility when infrastructure becomes interchangeable.
Data isn’t a byproduct of product usage — it is the moat.
Services Are the Software Now
In the age of AI, the traditional SaaS playbook — sell the software, let the customer figure it out — doesn’t cut it. AI tools often exceed the technical readiness of the teams adopting them. The result: a widening gap between product capabilities and customer outcomes.
AI doesn’t eliminate the need for services. It transforms them. The most successful startups today are embedding service-like functions directly into the product experience. Instead of expecting customers to learn and adapt, they are:
- Encoding onboarding and training into intelligent workflows.
- Guiding integration through AI-powered assistants.
- Offering in-product optimization that evolves over time with usage data.
This is not the same as running a classic services team. It’s a rethinking of delivery: designing products that perform the service, rather than requiring one.
Here’s the new formula for delivering value — and defensibility:
- Horizontal capability: A powerful toolset addressing critical worflows.
- Vertical implementation: Deep understanding of industry-specific or function-specific workflows and outcomes.
- Proprietary or private data: Feedback loops that improve performance over time.
Where traditional SaaS added services to patch gaps, AI-native companies collapse the boundary between product and service. For vertical AI startups, domain-specific implementation drives faster adoption. For horizontal platforms, services — embedded or adjacent — are the bridge from utility to indispensability.
The goal isn’t to monetize services. It’s to ensure outcomes.
When customers struggle to extract value on their own, smart companies step in — not with billable hours, but with automated onboarding, guided experiences, embedded expertise, and co-pilots that feel like part of the team.
Services, when designed intentionally, don’t just drive adoption — they generate the data, trust, and insight that fuel long-term product differentiation and customer retention.
What Founders Should Expect: Market Signals to Watch
Even in uncertain times, some things are getting clearer. Here’s what founders should expect in the months ahead:
- IPO window reopens, but selectively: Exceptional companies — think Figma-level breakout — will lead the way. Everyone else needs to stay focused on building durability and scale.
- SaaS consolidation is coming. Most tools won’t survive. The winners will consolidate categories or disappear.
- Fewer winners, deeper moats: In some categories, there may only be two real winners with large outcomes. If you’re third, speed and reinvention are your only lifelines.
- AI will replace services, not just software: The biggest opportunity isn’t to displace legacy tools. It’s to replace the high-cost human-driven services layered around those tools.
- Big companies aren’t sleeping giants anymore. For the first time in years, startups need to worry about incumbents getting there first. AI enables them to innovate faster than they ever have before.
The Next Generation is Just Getting Started
We’re only two years into the generative AI wave, and yet the innovation is relentless. The companies that define this decade are being built right now — not by chasing hype, but by building products that go deeper, embed tighter, and evolve faster than anything that came before.
Netscape Navigator launched in 1994. But 99% of the internet’s market cap today was built by companies that came two years after its release. ChatGPT hit the scene just two years ago. Let that timeline serve as a reminder: This is the beginning — not the end — of what’s possible.
If you’re a founders building the future with AI, please reach out to us at [email protected] and [email protected]. We’d love to partner with you.