Emerging Business Models in Enterprise Software – Part II
The new infrastructure of enterprise AI — personalized software, agents, and data platforms
Tarush Aggarwal
April 14, 2025
Last time, we explored how the SaaS playbook is breaking — cloud isn’t getting cheaper, software moats are shrinking, and AI is making development dramatically faster and cheaper. If you didnt read my previous article read Rising Costs & the Decline of SaaS
This time, let’s talk about what comes next.
We’re entering an era where companies adopt AI to accelerate business in ways unimaginable just a few years ago. And in this world, four categories of companies will thrive.
Think of them as the new “shovel sellers” of the AI gold rush — building the infrastructure and tooling for the next generation of business. And the winners won’t look anything like the SaaS giants of the last decade.
1. Personalized Software — The Evolution of SaaS (Layer 1)
We’re seeing the rebirth of software — this time, it’s personalized.
Why?
SaaS is getting expensive, fast
It’s now radically cheaper to build software using AI.
Enter a new wave of firms — part software, part consultancy. These companies specialize in a single category (CRM, ERP, helpdesk, etc.) and deliver tailored solutions using core templates and AI-powered customization. Think of them as the modern evolution of SI (system integrator) partners — but built for speed, scale, and specificity.
You get the best of both worlds:
From software: Continuous improvements, stability, credibility.
From services: Custom-fit experiences and domain-specific depth.
But without the usual downsides:
No bloated consulting headcount.
No one-size-fits-all UI that forces your team to adapt.
AI enables a leaner model — upfront implementation fees, ongoing maintenance, and pay-as-you-go customization. These products are community-driven, template-based, and increasingly self-evolving.
These companies will dominate single software categories and, for the first time, collapse the vendor + services partner into a single entity.
2. Agentic RAG — The Execution Layer (Layer 2)
On top of this personalized software stack sits a new kind of interface — one that interacts on your behalf.
We’re moving from using software via screens… to interacting through intent: voice, chat, and intelligent automation.
What began as “Hey Siri, call an Uber” is evolving fast: