Is AI Killing SaaS, or Just Exposing Weak Architecture?

HubSpot is down over 60% but the fundamentals are fine. The real story is architectural — and it changes how enterprises should choose their stack.

HubSpot is down more than 60 percent over the past year

At first glance, that looks like a broken company. Look closer and the fundamentals tell a different story. Revenue is still growing. Customers keep expanding. Retention is strong. Free cash flow has improved.

The business did not collapse. The market's belief about the business did.

The multiple collapsed, not the company

For most of the past decade, SaaS lived inside a clean narrative. Recurring revenue. High gross margins. Expanding wallet share. Platform lock-in. HubSpot, Salesforce, Atlassian — they all checked the same boxes.

Investors paid for that narrative. More precisely, they paid for the assumption that whatever platform owned the workflow would keep owning the value.

Then AI changed where the workflow actually lives.

AI introduced a new layer in the stack

The biggest shift of the past two years is conceptual, not technical.

AI now sits between the user and the software. Marketing teams used to write campaigns inside HubSpot. Today, they brief an agent that drafts the campaign and pushes it to the platform. Sales teams used to enter notes in Salesforce. Now a transcription agent does it after the call ends.

The platform is still there. But it has quietly been demoted — from workspace to database.

Re-thinking your stack for an AI-first world?

Let's talk through your architecture. We've helped enterprise teams escape monoliths and stay AI-ready.

Let us talk

That is what the market is repricing.

AI features are becoming the new normal

Every SaaS vendor has responded the same way: ship AI features. Copilots. Assistants. Smart recommendations.

HubSpot has ChatSpot. Salesforce has Einstein. Atlassian has Rovo. Optimizely has Opal.

This is no longer optional. Vendors who do not ship AI features will disappear. Vendors who do are reaching parity quickly. When every platform has a similar assistant, the assistant stops being a differentiator. It becomes the new floor.

A smarter button is not a real advantage when every competitor ships the same button next quarter.

The underrated piece: orchestration and human-in-the-loop

The real differentiator is one step deeper, and most coverage misses it.

AI without orchestration is a risk. An agent that generates content, qualifies leads, or triggers actions without proper guardrails, audit trails, and human review can move fast in the wrong direction. Errors scale at machine speed. Compliance teams notice. Customers notice.

The platforms quietly winning the next round are the ones investing in the unglamorous parts. Workflow governance. Approval flows. Versioning. Observability. The ability to keep a human in the loop without slowing the machine down.

These features rarely make the press release. They make the difference between AI that works in production and AI that creates expensive mistakes.

This is really an architecture question

A monolithic suite locks data, workflow, and UI into a single vendor. Whatever AI agent you want to use has to go through the suite's own assistant, on the suite's terms, with the suite's roadmap.

A composable, API-first platform is different. Its data is addressable. Its workflows are scriptable. Its UI is replaceable. Whatever AI layer wins this decade can sit on top of it without ripping anything out.

This moment is not really about AI versus SaaS. It is about closed versus open architectures — played out in public market valuations.

AI raises the floor, not the ceiling

There is a quieter side to this story that often gets missed.

AI does level the playing field — but only at the lower end. A junior marketer with a good AI tool now produces output that a year ago would have needed someone two levels more senior. Mediocre work gets faster. Generic work looks more polished.

That is real, and it is useful.

But excellence does not come from plugging into an AI assistant. The professionals who will pull ahead are the ones who use AI as an instrument, not a substitute. They bring judgment about what is worth doing, taste about what good looks like, and the discipline to keep humans in the loop where it matters.

The same logic applies to platforms and to teams. The winners will not be the ones who ship the most AI features fastest. They will be the ones who take orchestration and human judgment seriously.

What this means for buyers

There is a regional dimension worth naming. US markets tend to run ahead on AI optimism — HubSpot's rebound from its 2026 lows reflects part of that. Europe, and DACH especially, runs the opposite risk: lagging adoption, sometimes for good reasons (GDPR, NIS2), sometimes for less good ones. Neither extreme is right. The thoughtful position is to invest in the architecture that lets you adopt AI well — not fast, not fearfully, but soundly.

Most of our work at Bright Global involves helping enterprise teams move off legacy monoliths and onto composable stacks — Optimizely, Contentful, Storyblok, commercetools, Shopify, Medusa.

Two years ago, the case for that shift was largely about marketing speed and editor experience. Today, there is a second argument no digital leader should ignore. AI orchestration only works on architectures that AI can actually orchestrate. And AI orchestration is only safe when humans stay in the loop.

That is a strategic difference. Not a feature difference.

The harder question the market is really asking

The HubSpot selloff is not a verdict on HubSpot. It is the market quietly asking a harder question.

In an AI-first stack, who owns the workflow — the platform underneath, the agent on top, or the human keeping both honest?

The companies best positioned to thrive — vendors and buyers alike — share two traits. Their architecture stayed open. And they treat AI as something to orchestrate, not something to outsource judgment to.

That is the advantage worth building right now.

What still builds a real advantage in an AI-first stack

  • Proprietary data depth

    Data no one else has, structured well enough for AI to reason over.

  • Workflow specificity

    Being the system of record for a workflow AI alone cannot fully replace.

  • Orchestration and human oversight

    Keeping humans in the loop as AI scales decisions across teams.

  • API-first openness

    Being easy to compose with is what keeps you relevant when the orchestration layer changes.

Bright Global
  • Expertise
  • Work
  • Approach
  • Partners
Bright IT
  • Expertise Overview
  • Architecture Audits
  • Legacy Liberation
  • Enterprise Platforms
  • B2C Commerce
  • UX/UI Design
  • B2B Commerce
  • Integrations & Middleware
  • AI & Automation
  • Continuous Operations
  • References & Industries
  • HDI Global
  • Swarovski Optik
  • Mischek
  • Dachstein Salzkammergut
  • Salzburg AG
  • Tyrolit
  • Vaillant
  • Our Philosophy
  • Direct-to-Expert
  • Nearshoring 2.0
  • Security & Compliance
  • Partner Overview
  • Storyblok
  • Optimizely
  • Contentful
  • Shopify
  • Medusa
  • commercetools
  • Ecosystem Integrations
Insights
Contact
Bright IT
Insights
  • Legal Notes
  • Privacy Policy
  • Follow us on LinkedIn
© 2026

Partners