“At Klarna, we’re transforming the way we collaborate with our GenAI chatbot Kiki, powered by Neo4j’s knowledge graph”

At Klarna, innovation thrives on experimentation. When generative AI (GenAI) began to emerge as a transformative force, Klarna didn’t dictate how it should be used — it encouraged employees to explore its potential on their own terms. 

“At Klarna, we decided early to explore the potential of AI and LLMs—mostly ChatGPT—while being open to testing all things that seemed to be trending,” said Sebastian Siemiatkowski, Co-Founder and CEO of Klarna in a post on X. “We encouraged all employees to do so and allowed them to pursue ideas organically rather than following “management direction” on exactly what they should be building.”

This organic approach sparked a wave of creative problem-solving that led to one of the company’s most significant breakthroughs, according to Klarna’s blog: Kiki, an AI-powered internal assistant built on Neo4j knowledge graph technology, and integrated with OpenAI’s large language models (LLMs). It was during this period of experimentation that Klarna discovered a significant challenge: data fragmentation.

“Enterprise software has a standard set of features that are vital for it to operate—features such as audit, versioning, access and edit management, and similar universal needs,” Siemiatkowski posted on X. “We need them as well, but that fragmentation again adds friction, admin overhead, and more.”

“In the early days of ChatGPT, we heard a lot: ‘this tool allows you to feed all your PDFs, all your data sources to a LLM!'” Siemiatkowski added. “However, the old universal truth of data scientists still holds true, even in AI: ‘sh*t in, sh*t out.”

Source: Sebastian Siemiatkowski on X

This realization led Klarna to re-examine how its organizational knowledge was shared.

“We started obsessing about the concept of collaboration on information,” Siemiatkowski explained in Sequoia Capital’s Training Data Podcast. “We’ve looked a lot to Wikipedia and other knowledge graphs, and how people collaborate on building great information.”

Why “Feeding an LLM the fractioned, fragmented, and dispersed world of corporate data will result in a very confused LLM”

Source: Sebastian Siemiatkoski on X

Klarna noted that its existing information architecture consisted of thousands of SaaS applications, from Salesforce and Workday to various project management tools. Analyzing team structures, understanding processes, or answering cross-functional questions required accessing multiple systems.

Source: Sebastian Siemiatkowski on X

Klarna began by constructing a knowledge graph of its data sources. 

“We started consolidating; put things together, connect our knowledge, remove the silos,” says Siemiatkowski on X. “The side consequence of this: liquidation of SaaS—not [sic], but a lot of them. Not for the license fees, though those savings are nice, but for unification and standardization of knowledge=data.”

Klarna then integrated OpenAI’s large language models (LLMs) to make the knowledge graph accessible to all employees, according to Klarna’s company blog — an integration that allowed employees to query the knowledge graph using natural language through a chatbot interface available in Slack and on Klarna’s internal knowledge platform.

“As Klarna continues to discover applications for OpenAI’s tech, there’s the potential to take the business to new heights,” says Siemiatkowski. “We’re aimed at achieving a new level of employee empowerment, enhancing both our team’s performance and the customer experience.”

When an employee needs to understand how teams are structured or find a specific process document, Kiki doesn’t just return isolated data points. It shows how information connects to everything around it—linking people to projects, documents to teams, and processes to outcomes. This approach means employees can see not just what they asked for, but also the surrounding context that makes the information useful.

Kiki Answers 2,000 Questions Per Day

Klarna’s blog says that Kiki “significantly aids in the management and distribution of internal knowledge, reinforcing Klarna’s commitment to a culture of transparency and open information flow.” What began as an experiment has evolved into an “internal AI assistant, which has adeptly responded to over 250,000 inquiries (2,000 per day) since launching in June 2023.”

“Beyond merely fetching information, Kiki cultivates a self-service and autonomous culture, empowering employees to promptly find answers and solve issues independently,” according to Klarna. “This not only boosts productivity but also reduces the time dedicated to administrative tasks, freeing up Klarna employees to concentrate on strategic and creative tasks.” 

The impact extends far beyond simple information retrieval. According to Klarna:

• Adoption: 85% of Klarna employees actively use Kiki, with non-technical teams seeing adoption rates as high as 93% in Communications, 88% in Marketing, and 86% in Legal.
• Simplification: Klarna has deprecated over 1,200 SaaS systems including Salesforce (CRM) and Workday.
• Productivity: Employee questions are answered within 1-5 seconds and dramatically improving collaboration

“Kiki brings together information across multiple disparate and siloed systems, improves the quality of that information, and explores it, enabling our teams to ask Kiki anything from resource needs to internal processes to how teams should work,” Siemiatkowski shared with Neo4j. “It’s having a huge impact on productivity in ways that were not possible to imagine before without graph and Neo4j.”

Get in Touch

Curious about what insights you could unlock for your business with graph-powered solutions? Let’s talk — reach out, and we’ll get in touch.