Klarna's AI pivot and the emerging Intelligent Enterprise

When Klarna's AI assistant began bypassing Salesforce and handling two-thirds of customer inquiries in its first month, I had to investigate to find out what it means for the SaaS industry. 

If you missed the story, Klarna’s AI-powered customer service assistant, built on OpenAI’s ChatGPT framework, handled two-thirds of inquiries within its first month, reducing average resolution times from 11 minutes to 2 minutes. This contributed to an estimated $40 million in profit improvements in 2024. However, CEO Sebastian Siemiatkowski later clarified that the company’s 50% workforce reduction (from 3,800 to 2,000 employees) was not purely a cost-cutting measure but part of a broader initiative to “do much more with less” through AI-human collaboration.

Contrary to initial reports, Klarna did not fully replace Salesforce with AI. Instead, it consolidated fragmented SaaS tools (including Salesforce CRM and Workday HR) into an internal tech stack using Neo4j graph databases to unify data silos. 

Turns out, Klarna wasn't declaring war on SaaS. They were fighting something far more insidious.

Fragmented knowledge.

His team discovered that enterprise knowledge was fragmented across countless systems. Each with their own data models, interfaces and access controls. A reality most enterprises face today. 

The average enterprise now uses over 125 separate SaaS applications.

This sprawl isn't just expensive and inefficient, it actively stops innovation and creates what Siemiatkowski calls "an unnavigable web of knowledge that required a tremendous amount of Klarna-specific expertise to operate."

Rather than feeding fragmented corporate data into AI systems (resulting in what he candidly describes as "a very confused LLM"), Klarna built something remarkable. 

A unified knowledge graph using Neo4j that connected previously siloed info.

This consolidation wasn't primarily about cost savings, though shutting down approximately 1,200 SaaS tools certainly helped their bottom line. 

It was about breaking down barriers between their data on customers, employees, products and operations.

And it looks like Enterprises that thrive in the AI era won't resist adoption but will solve the critical challenge of fragmented org knowledge. 

As Siemiatkowski puts it: "Opinionated software is worth something, as opinions represent an experience of what works, what produces results."

The most forward-thinking companies aren't simply automating existing processes. 

They're reimagining what's possible when humans and AI work together on newly unified data foundations.

Which raises the question: 

How much productivity is your org losing to fragmented knowledge across dozens of disconnected SaaS tools?

Klarna blog: https://lnkd.in/ghUgXQkm