20VC: SaaS is Dead: Why Systems of Record Will Die in an Agentic World | What Revenue Multiple Will Software Companies Trade At? | From 7,000 to 3,000: We Need Less People Than Ever with Sebastian Siemiatkowski

What It Covers

Sebastian Siemiatkowski, Klarna CEO, explains how AI enabled his company to shrink from 7,000 to under 3,000 employees while expanding services, why traditional SaaS businesses face existential threats from falling switching costs, and his prediction that software companies will trade at utility-like valuations as AI makes code generation nearly free.

Key Questions Answered

• •Headcount reduction through AI: Klarna decreased from over 7,000 to under 3,000 employees through natural attrition of approximately 20% annually, while simultaneously launching new banking products without requesting additional budget. Employee compensation per head increased nearly 50% during this period, sharing productivity gains with remaining staff to maintain morale and retention during the transformation.
• •SaaS valuation compression: Software companies historically traded at 20-30x price-to-sales ratios but have fallen to 5-10x, with potential to drop to 1-2x like utilities. Chegg trades at 0.2x after ChatGPT disruption. The fundamental shift occurs when AI agents reduce data switching costs, eliminating the moat that justified premium valuations for enterprise software companies.
• •Customer service transformation model: Klarna built an Uber-style model recruiting passionate customers in rural areas for part-time customer service work. These customer-agents demonstrate superior NPS scores because they genuinely use and understand the product. This approach combines AI handling simple queries with human relationship-building for VIP experiences, recognizing artisan human connection becomes premium in an AI-automated world.
• •Enterprise data compression thesis: AI functions as compression technology, storing information once rather than duplicating across Slack, Salesforce, Google Docs, and other systems. Wikipedia demonstrates this principle with one article per topic versus enterprise duplication. ChatGPT-5 equivalent model size equals just three days of global weather data, suggesting enterprise compute needs may dramatically decrease as organizations eliminate redundant information storage.
• •Banking product expansion strategy: Klarna converted 30 million US buy-now-pay-later users into full banking customers, reaching 2-3 million active cardholders within months. The company removed revolving credit features, sacrificing $100 million in revenue, to offer fixed installment payments as a healthier alternative to traditional credit cards. Twenty percent of transactions now process as debit, reintroducing the choice banks previously eliminated.
• •AI development positioning: Anthropic's Claude optimizes for intelligent advisory relationships and unbiased feedback, while OpenAI's ChatGPT evolves toward emotional companion experiences with higher engagement metrics. Enterprise customers require AI that challenges assumptions rather than pleasing users. This divergence creates distinct market positions, with Claude serving professional contexts and ChatGPT targeting consumer emotional connection and entertainment use cases.