There’s never been a better time to be in product.
Cursor, Lovable, and every co-pilot clone you can think of have turned building software into a game of prompts. AI is compressing development time, amplifying individual contributors, and giving product teams a runway unlike anything we’ve seen before. If you can say it, you can ship it.
Venture capitalists, ever the fans of repeatable models, are responding accordingly.
They’re underwriting a tsunami of point solutions that follow the classic SaaS playbook. Find a niche, identify a discrete problem, build a product with a tight feedback loop and scale until your ARR hits the roof.
YC’s latest cohort reads like a menu of painkiller micro-SaaS with four “Cursor for X” bets that take the pattern-matching approach to new heights. We see the same trend ripple across the entire investosphere. According to EY, VC-backed companies raised over $80 billion in Q1 2025 alone, with IT and software leading the charge.
Who could blame them? Thanks to AI, it’s easier than ever to launch products that have wide enterprise appeal.
And yet.
Underneath the sea of point solutions, a much deeper shift is brewing. Driven by client preferences and the simple tenets of contract theory, the age of AI is ushering in nothing less than the great servicification where product-led-growth gives way to an outcomes-obsessed model.
The Quiet Collapse of the Product
The economic data already pointed the way.
Services make up over 72% of U.S. GDP. Globalisation, digitisation, and labour market evolution have all conspired to favour services over products in the grand scheme of the economy. In software, the first wave of servicification came with SaaS, but look closely and the mirage of services is dispelled, revealing underneath it nothing more than a billing model change. When Adobe ended perpetual licenses for Creative Suite in 2013 and moved to Creative Cloud, it didn’t change the product, only how you paid for it. Microsoft Office followed suit, with the entire industry following soon enough.
Behind the SaaS, we’re still paying for tools users have to learn, manage, and operate, just now the product is rented monthly. SaaS is service in name, product in nature, all-round ARR-machine.
With AI, the landscape is changing rapidly. This time, the subscription model is sticking and it’s the product that is vanishing.
As Gautam Ajjrapu, co-founder of Glide, puts it: “You still need a killer product to earn your seat at the table. For us, that meant rebuilding onboarding flows so cleanly that banks said: ‘Now help us fix retention, too.’”
The product remains the credential. But what earns the seat at the table is the service logic beneath it.
“We’re not shipping dashboards or generating metrics. We’re delivering a clear outcome by bringing customers through the door. Now, we’re shadowing loan officers, mapping every manual workflow, and building AI agents to do that work. What we’re building is a service nestled inside a product wrapper,” Ajjrapu continues.
In Glide’s case banks may start by looking for tooling, but what they stay for is the transformation in how they reach their outcomes.
“Every time a bank says yes to Glide they’re buying the promise that someone finally understands their workflows, and will quietly erase the mess behind them,” Ajjrapu adds.
At a time where “evolve or die” has become standard operating procedure, the transition from selling products to selling survival is what has kicked Glide into high gear.
“We didn’t lean on legacy credentials to get started, and we know our Gen Z background makes us outliers. That’s also exactly what made us useful, because we’re building the system that replaces the old one, not fixes it.”
The product opens the door, but the outcome is what keeps it open.
The Agent is the Service
Look at the most hyped AI companies of 2025: Agentforce, EMA, and OpenAI’s operator with too many competitors to count. They still sell licenses, but the form factor around their offering is radically different. Yes, each company still sells tools that help you do work. However, the more we progress on the agentic spectrum towards autonomy and systems-level integration, the more these companies sell outcomes that happen for you. Just like a service company would.
Swipe a card, and an agent triages your inbox, preps your investor memo, or files your expenses. The UI itself is disappearing and being replaced by API calls and DM’s on Slack. The workflows once etched into Visio maps are disappearing, obfuscated by the agentic action that might as well be invisible even if it was fully observable. What remains is the result.
Look at Zendesk for example, which now sells “automated customer service” with AI resolving tickets end-to-end. Amplitude just launched its fleet of AI agents that not only interpret product data but generate growth recommendations and even implement A/B tests for you. These aren’t products in any traditional sense. These are outright services, purchased through the viewport of a product.
As opposed to many other facets of the AI transition, the shift toward outcome-driven delivery is more than just branding and marketing. Instead, we’re seeing a structural realignment of how value is defined and delivered to the customer.
This concept is particularly resonant in healthcare where the only thing that matters is delivering better care.
“Doctors don’t need more software,” says Andreas Cleve, CEO of Corti. “They need more time to treat patients. If technology gives them that, they don’t care what it looks like under the hood. Neither do the patients they treat”
That’s the core of AI-driven service delivery, getting to outcomes in ways that feel less like using a product and more like gaining capacity.
“The moment a tool becomes invisible and just works, that’s when we’ve done our job,” Andreas continues. “The best systems in healthcare are the ones nobody notices because the outcomes speak for themselves.”
What companies like Corti are realizing is that while AI is the path to getting these outcomes at scale, general-purpose models simply won’t cut it. Instead, what is required is something far more powerful, and often more specific.
“Healthcare doesn’t run on guesses. You need models trained on the right data, tuned to real workflows, and able to hold up under clinical pressure. That’s where niche models and AI agents come in,” Andreas explains.
Agentic autonomy layered on high quality data is making the servicification possible, as if not entirely inevitable.
“You can’t sell point solutions where others are selling outcomes,” Andreas argues. “What we’ll inevitably see are more and more software companies taking their clients further and further to the outcomes, to the point where the product itself becomes invisible.”
Contract Theory and The Inevitability of Services-As-A-Service
The future tends to look a lot like the past. Only this time, there’s a promise of greater margins.
Think of your friendly neighborhood law office and how they’ve set up themselves to focus on only what matters. No partner wants to worry about maintaining the office printer, they just want documents printed when needed which is why they’ve all outsourced it. The brand of the printer and its specs matter much less than the output and the ease with which they can be accessed.
Nobel Prize winner Bengt Holmström’s contract theory captures this perfectly. On the most fundamental level, companies want to buy results, not tools. No one truly buys a CRM in order to configure Salesforce workflows, they buy it to drive more sales. No one wants to interpret an FP&A dashboard for the pure joy of it, they want forecasts they can act on. The rise of AI is simply extending the inevitable logical outcomes of contract theory to more domains, stripping away the middleware between intention and outcome.
In the old days, the desire for outputs birthed the action-oriented consultancies that sold strategic outcomes, not just the tools. Now, tech companies are converging on the same model. Instead of a product that demands onboarding, configuration, and constant ops management alongside an annually increasing subscription, they are on the path to offering set-it-and-forget-it AI agents that simply get the job done.
Here, the product becomes secondary to the outcome. Whatever AI model, patented tech stack or integration process you offer, the client is in it only for the outcomes that are enabled by the tech. Call this Service-as-a Service, if you will.
It starts subtly enough. First Gong offers sales analytics. Then Gong adds suggested scripts. Then Gong reps deliver training. Then Gong sells AI that listens, learns, and nudges your team mid-call. Is there any reason to believe that at some point, the offering stops being a product and becomes an AI-powered sales consultant entirely. The delivery is automated and the process is de-emphasized, but the outcome is closer to what the client is truly searching for.
“A lot of consultants can map the problem, but they can’t build the fix,” says Gautam Ajjrapu, CEO of Glide. “We asked what if the partner who diagnoses the issue is also the one who engineers the solution? That’s what banks want, more work done, more clients served.”
Some like Glide and Corti are already delivering on this, while others are still wrapping a tool in layers of promises.
The Fall of Product Led Growth Is Near
Why is this happening now with AI and not before?
Yes, customers have always preferred outcomes over infrastructure, but just finally built the tech that can deliver them without the bloat in between.
Ask any ops leader if they’d rather implement a new system or just hit their quarterly target and you’ll see what I mean. Any leader worth their salt will choose the latter every time, given enough control and trust in the vendor and what they offer. But until now, you needed humans to do the work.
With agentic AI, you can automate out the messy middle and go straight to the good stuff.
Want better sales emails? Don’t buy a product to help you write them. Pay an AI service to qualify the leads, send the emails and convert the customers into an automated pipeline. Want onboarding fixed? Don’t buy a LMS. Hire an AI-first startup that uses agents to connect everything from the job description to the first salary-slip.
What’s driving the VC conviction in all of their newly funded products is how AI makes them scalable. What most of them are missing is how we’ll all find the very idea of a standalone product quaint, simply because of how customers will choose a competitively priced outcome over a product each and every time.
To be clear, the fall of the product is neither immediate nor universal.
Case in point being how fax machines still linger in plenty of government offices, and there will always be laggards in the corporate realm as well.
But look closely, and you’ll see the pioneers moving along.
“Most systems today give you fragments,” says Mel Morris, CEO of Corpora. “We built ours to give clients synthesis and outputs instead of simply pushing them along a few steps in the process.”
“Corpora’s idea is not to replace research,” Morris adds, “but to compress it, structure it, and return something actionable. If that happens, users stop thinking about the product entirely, they just simply take its outputs for granted and build on them.”
Elsewhere, we’re seeing Glean has expanded from enterprise search into a knowledge ops function that delivers answers, not just documents. ElevenLabs began by cloning voices and is now providing brand voices-as-a-service. The go-to-market across the industry is shifting, too. More and more vendors are bundling advisory, onboarding, and implementation not as add-ons but as core value.
“People talk about product-market fit,” Morris says, “but what really matters now is output-context fit. If you deliver clarity at speed, that’s what wins, not the wrapper it comes in.”
What might still look like product-led growth is actually outcome-led delivery, with the product as the excuse to get started.
If you’re building AI tools today, ask yourself: are you really a product company? Or are you a services firm in disguise?
The real opportunity is to lean in to the answer that goes against the advice that gets you into the annals of YC. Don’t just build scalable point solutions for users. Build outcomes that scale from one client to the other. Design with the obsolescence of the interface front and center, and make your product the conductor, not the first violin.
Because maybe, just maybe, Service-as-a-service will be the new SaaS simply because the outcomes are what customers have always wanted. They just didn’t know it was possible, until now.