.png)
Peak software? AI agents, Claude Code, and SaaS selloff signal a shift in software economics. Explore AI’s impact on SaaS, HCM, pricing models, valuation, and the future of agent-first platforms.
February delivered a jolt to the system.
Between the rapid rise of Claude Code, bold predictions that 95% of software code will be written by AI within a few years, and a public market selloff that punished anything labeled “SaaS,” it feels like the ground shifted underneath the software industry almost overnight.
The common thread in all of it: a fundamental input to the SaaS business model has changed forever.
Software development, once expensive, slow, and scarce, is becoming abundant, fast, and nearly free. That single shift has massive downstream implications for valuation, competition, pricing, and power dynamics across the entire software landscape.
Which leads to the question everyone is asking, sometimes quietly, sometimes loudly:
Many investors and commentators rushed to declare that software is finished. That the golden era is over. That agents will replace SaaS entirely.
That conclusion feels premature.
Software isn’t dead. But I do believe we’ve reached peak software.
We’re no longer on the front nine of the SaaS era. We’re making the turn.
Like railroads, automobiles, or telecom before it, software has taken it's turn as the dominant industry in America. It reshaped how companies operate, scale, and compete. And like every dominant model before it, it won’t stay dominant forever.
What we’re entering now is a transition period where agent-first models begin by layering on top of SaaS, then gradually eat into it. Over time, many traditional interfaces and feature sets will dissolve into workflows, conversations, and outcomes rather than screens and modules.
That doesn’t mean everything collapses tomorrow. Business cycles move slowly. Enterprises move even slower.
Which brings us to the real questions worth asking.
If software isn’t dead yet, then:
I’ve been fairly consistent here, even when headlines get breathless.
In the short term, the impact of AI will be smaller than the hype suggests. In the long term, it will be much bigger than we can fully imagine.
Over the next 1-3 years, I expect the HR Tech space to look largely the same.
The major players will remain the major players. Growth may be choppier as CIOs and their AI budgets pull spend away from traditional HR projects, and companies are hesitant to make big system changes amid an unsettled landscape.
But it will take time for AI-first tools and workflows to trickle down.
Just last week, I was on a call with a buyer asking about PeopleSoft and Taleo. More than two decades after cloud-native platforms reshaped the market, some buyers are still asking about on-prem platforms.
The more meaningful changes will emerge over a 4-7 year horizon.
That’s when I expect AI-native platforms to begin achieving real scale and product-market fit, meaningfully impacting today's household software brands.
It’s also when Google, Microsoft, and OpenAI agents start increasingly encroaching on workflows that historically lived inside HCM, ERP, and CRM platforms.
People forget that Google once had an ATS. Microsoft already has Viva, their employee experience platform. These efforts stalled in a pre-agent world. They won’t stall this time, because development costs are near zero and agents are increasingly context-aware across teams, workflows, and data.
I think you’ll also see a new type of competitor emerge: small, bootstrapped agencies that build custom agent layers and internal tools directly for businesses. In this era, the “startup” won’t be a software platform, but an orchestration company stitching together agents, workflows, and internal systems.
All of that becomes a real headwind for traditional HCM vendors.
But here’s the important part.
Despite all the disruption, HCM has a massive structural advantage.
Traditional software is deterministic. Large language models are probabilistic by design. They are powerful, flexible, and fast, but they are not consistently precise.
And in certain domains, precision matters more than efficiency.
When companies decide whether to a.) rely on an AI model, b.) vibe code a tool internally, or c.) purchase software from a vendor, the real tradeoff becomes liability versus efficiency.
That’s where categories like HCM, ERP and EMRs will still shine. The cost of being wrong is simply too high.
In fact, it’s not crazy to imagine large software vendors lobbying for more regulation, not less. In a world where compliance, trust, and accountability matter, regulation becomes a moat.
This is very good news for the biggest HCM platforms that handle payroll, tax, and compliance as core competencies.
It’s less promising further down the stack.
Mid-market best-of-breed tools have long differentiated themselves by having more robust feature sets than their all-in-one counterparts. When features can be copied overnight or generated on demand, vertical focus becomes less of an advantage.
I expect many popular point solution companies to shift toward more bespoke, services-heavy models, helping customers design and implement custom workflows rather than selling one-size-fits-all platforms.
The long tail of small, venture-backed point solutions will also take a big hit, squeezed from both sides by hyperscalers on one end and by companies' ability to generate their own tools on the other.
Several shifts feel inevitable.
1.) Marketing becomes a company's most important competency. When the 'product' portion of product-market fit is rapidly being commoditized, the marketing portion becomes even more critical.
Humans will still make big purchasing decisions, and increasingly, they will buy through their networks. Companies that can tap into channel partnerships and network effects will be the biggest winners.
2.) The value of features goes to zero. Buyers won’t care what a product does today if they can ask it to do something tomorrow and have it built instantly. Decisions will hinge more on philosophy instead.
How do you price usage?
How do you protect our data?
How do you interoperate with other platforms?
What design & build services do you wrap around the technology?
3.) UI becomes far less important. For better or worse, UI has driven many buying decisions historically. But as work increasingly moves into chat windows and eventually into voice interfaces, the value of UI will decline.
Part of the reason great UI's mattered was that they made it easier for users to accomplish tasks. If agents are the ones running tasks, then look-and-feel becomes irrelevant.
4.) The pricing models break. Per-seat pricing becomes harder to justify as agents do more work. Expect pricing tied to outcomes, jobs-to-be-done, or consumption. The biggest pressure will hit upsells, where companies used to charge for new features, analytics, and insights that orgs can now spin up on their own.
HCM is relatively insulated, but not immune.
Pressure will come from everywhere: hyperscalers, internal tools, agencies, and price compression.
The companies that survive and thrive will be the ones that get very clear on where they offer a liability advantage over efficiency-first alternatives.
They’ll move their experience toward chat and voice early, because that’s where the competition is heading.
They’ll embrace customization and composability, either directly or through partners, rather than fighting it.
And they’ll accept that we’re entering a slower-growth era that demands harder decisions.
Some incredible companies will emerge stronger, just like Target and Walmart did in the shadow of Amazon. But there will also be plenty of Toys “R” Us and K-Marts along the way.
Peak software doesn’t mean the end of software.
But, it does mean the easy years are over.
