If you run a Shopify store, sell on Amazon, or live inside TikTok Shop and Reels, you have probably felt the same tension investors are feeling about SaaS right now.
AI agents are getting good fast. The fear is simple: “If AI can build anything, why keep paying for tools?”
The more useful answer is also simple: AI will absolutely let teams build more custom workflows. But replacing real, production-grade software is still hard. And in ecommerce content, “hard” shows up in very specific places: brand safety, approvals, product truth, channel specs, and performance measurement.
So no, AI is not going to wipe out SaaS. But it will force changes in what “good software” looks like, and what you should build in-house versus buy.
The investor fear: agentic AI means companies can build bespoke software
In the SaaS world, stocks have been hit because investors are worried about agentic AI, meaning AI-powered software that can autonomously perform tasks.
The theory goes like this:
- AI coding tools keep improving
- Companies can “just build” what they need
- SaaS gets replaced by internal tools
- SaaS growth slows, so valuations fall
That story has a kernel of truth. Teams can build more than they used to.
But it skips the part that matters in the real world: enterprise-grade software is not just “code that works.”
It is security, compliance, identity and access, observability (knowing what broke and why), integrations, uptime, and long-term maintenance.
That is why companies pay for SaaS in the first place.
Two things can be true at once:
- You can build your own software now more easily than before
- It is still extremely hard to do well, at scale, for years
We have seen this movie before: cloud did not kill IT, it changed IT
Early in the cloud era, many IT teams resisted moving off on-prem systems.
Then the shift happened anyway. Resistance did not stop it. It delayed it.
The same pattern is forming with AI.
Tien Tzuo (Zuora founder, and early Salesforce employee) put it bluntly: every software company is facing an existential threat, and the only move is to meet it head-on and navigate through it.
That does not mean SaaS disappears. It means categories get reshaped, and some vendors lose if they do not adapt.
What the numbers say: big SaaS is not collapsing
A useful reality check from the reference article: look at the fundamentals of major enterprise SaaS companies that have embraced AI.
Across Salesforce, ServiceNow, Adobe, Workday, and SAP, recent quarters show something important:
- Growth is mostly steady, not falling off a cliff
- AI-related revenue or bookings are showing up and scaling
- In some cases, guidance is being raised or expectations are being beaten
Salesforce, for example, showed revenue growth accelerating across recent quarters, alongside growth in its data and AI products.
ServiceNow’s “Now Assist” hit $600M.
Adobe reported that “AI-influenced ARR” is now more than one-third of its overall book of business, with “AI-first ARR” crossing $250M.
Workday reported strong AI attachment in deals and over 1 billion AI actions on its platform in a year.
SAP’s growth was steady, and it argued that AI does not automatically mean “everyone can code their own SAP.”
If “vibe coding” was already eating core SaaS categories like CRM, ERP, and HR, you would expect to see revenue impacts by now. We are not seeing that.
Why this matters to ecommerce operators (not just Wall Street)
Ecommerce teams are not debating CRM replacement. You are debating something more immediate:
“Can we stop paying for creative tools, agencies, and production workflows because AI can generate videos now?”
That is the content version of the same SaaS fear.
And the answer is similar:
You can generate more content in-house than ever.
But running content like a production system is still hard.
Here is what “enterprise-grade” means in ecommerce video:
- Brand consistency across hundreds of assets
- Claims compliance (especially beauty, wellness, supplements, finance)
- Rights management (music, creator likeness, usage windows)
- Channel-specific formatting (TikTok, Reels, Shorts, Amazon, PDP video)
- Localization without breaking meaning
- Measurement and iteration tied to conversion, not vibes
- A clean pipeline from product data to creative output
AI helps with the making. It does not magically solve the operating.
The three SaaS fears, translated into AI video and social commerce
The reference article highlights two big investor concerns and one underlying reality: data is the roadblock.
Let’s translate that into your world.
1) “AI coding tools are rapidly improving” becomes “AI video tools are rapidly improving”
Yes.
You can now spin up:
- UGC-style scripts in minutes
- 20 hooks for the same product
- Variations of the same ad for different audiences
- New edits from existing footage
- Product showcases from a feed of images and specs
This is real. It is why content volume is exploding.
But the hard part is not generating one video. It is generating 200 videos that are all on-brand, accurate, and measurable.
2) “Companies want fewer humans” becomes “teams want more output without more headcount”
Also yes.
Most brands are not hiring a bigger creative team just because TikTok Shop wants daily posting and Amazon wants better video on listings.
So the pressure shifts to systems:
- templates
- repeatable workflows
- approvals
- performance feedback loops
This is where “SaaS vs build” becomes “workflow vs chaos.”
3) The hidden constraint: you cannot vibe code the data
This is the most important part of the reference article, and it maps perfectly to ecommerce content.
SAP’s CEO explained the roadblock customers hit when they try to build custom agents: LLMs are great with unstructured text, but the agent still needs the business data. The payment info. The pipeline. The context.
In ecommerce video, your “business data” is:
- product catalog truth (titles, variants, ingredients, materials, sizing)
- pricing and promos by channel
- inventory status
- shipping promises by region
- review themes and objections
- return reasons
- creative performance history (what hooks worked, what failed)
You can generate endless videos. But if the videos are not grounded in the right product and performance data, you get:
- wrong claims
- wrong variant shown
- wrong price
- wrong offer
- wrong audience angle
- inconsistent brand tone
That is why the future is not “one person with an AI tool.” It is AI plus a data-connected content system.
This is also why big SaaS companies have an advantage: they sit on oceans of data and can build AI on top of aggregated history, not just one company’s isolated dataset.
Ecommerce brands feel the same gap. Your store has a pond of data. Platforms and large tools often have broader patterns.
The winning move is not pretending you can replace everything. It is building a stack that lets your data actually drive your creative.
What changes in practice for Shopify, Amazon, TikTok Shop, and Meta commerce
AI will not remove tools. It will change what you demand from them.
Here is how that shows up by channel.
Shopify merchants: PDP video becomes a system, not a one-off project
On Shopify, video is increasingly part of conversion, not just awareness.
AI makes it cheap to create:
- product explainers
- “how it works” demos
- comparison clips
- objection-handling videos (shipping, sizing, durability)
- seasonal refreshes
The change: you stop treating PDP video like a quarterly production sprint.
You treat it like inventory. Something you refresh, test, and improve continuously.
If you want the bigger picture on how AI is reshaping discovery and traffic, this connects directly to Google’s AI-driven search experiences and what that means for stores:
Google AI Mode Traffic for Shopify Stores
Amazon sellers: listing video and Sponsored Brands video need fast iteration
Amazon is ruthless about clarity and trust.
AI helps you produce more variations, but Amazon content has constraints:
- strict claims and category rules
- shoppers who want proof fast
- creatives that must match the listing exactly
The change: your advantage becomes iteration speed without breaking accuracy.
That means your video workflow must be tightly tied to listing data and review insights, not just “make it look cool.”