The founders productize the playbook they ran inside Klarna, Voi, and iZettle. The market they are launching into has already moved.
What does enterprise software look like when AI custom-builds itself into every workflow?
That question stopped being academic in February 2026. Roughly $285 billion in SaaS market capitalization evaporated in what analysts have started calling the SaaSpocalypse. Bain & Company published a report concluding per-seat pricing was structurally vulnerable to AI agent adoption. Monday.com replaced its 100-person sales development team with AI agents. The median revenue multiple for software firms fell from above 7 to below 5. Retool’s 2026 enterprise survey found that 35% of teams have already replaced at least one SaaS tool with a custom build, and 78% expect to build more in 2026.
Pit, the Stockholm-based platform launching today from stealth with $16 million led by Andreessen Horowitz, is built around a single bet. The bet is that the answer to the post-SaaS question is not another SaaS product. It is an AI-native infrastructure layer that produces real, production-grade custom software for each enterprise’s actual workflows, in days rather than months. The round includes participation from Lakestar, the Stena and Lundin families, and a dense angel list spanning founders and executives from OpenAI, Anthropic, Google, Deel, Revolut, Permira, DST, Goldman Sachs, Balderton, and Index.
https://www.youtube.com/watch?v=B32w47FZMec&embedable=true
The $1 Trillion Workflow Problem
Enterprise digital transformation has consumed over $1 trillion in spending since 2020. Most of the workflow problem it was meant to solve remains unsolved. Across industries, core operations still run on a fragmented mix of spreadsheets, email threads, rigid SaaS tools forced into shapes they were never designed for, and the brittle internal scripts engineers wrote to hold the rest together.
The reason is structural. SaaS products are built for the median customer. Enterprises run operations that are anything but median. Every company has its own approval chains, its own data taxonomy, its own compliance constraints, its own customer segmentation logic. The standard solution has been to either bend operations around the SaaS product’s assumptions or pay an integrator to bridge the gap with custom development. Both options are expensive, both are slow, and both produce systems that age into the next round of technical debt.
The Klarna Playbook
The reason Pit’s founders have credibility on this thesis is not the press release. It is what Klarna documented publicly across 2024 and 2025. Klarna disclosed that 90% of its employees use generative AI tools daily, with adoption rates of 93% in communications, 88% in marketing, 86% in legal, and similarly elevated rates across engineering, operations, finance, HR, and analytics. The deployment was not a single chatbot. It was custom AI software built into the specific shape of each function’s workflow.
Klarna’s customer service AI got the headlines, including the well-publicized partial reversal of that specific deployment in 2025. The broader pattern was less dramatic and more durable. The custom AI systems built into legal contract review, marketing campaign execution, communications drafting, and operations always outperformed the off-the-shelf SaaS tools they replaced. They took months to build. They required engineering teams Klarna had and most companies do not.
Why “AI Product Team as a Service” Is the Category Frame
Pit’s positioning is precise. It is not a low-code platform. It is not an AI copilot. It is not a vibe-coding tool. The company describes itself as an “AI product team as a service,” a framing that maps to a specific gap in the current AI tooling market.
GitHub Copilot and Cursor make individual engineers faster. They do not produce production systems on their own. Lovable, v0, Bolt and the broader vibe-coding category ship prototypes in hours, but the output is rarely the kind of thing a Fortune 500 company runs payroll or compliance on. Traditional engineering teams produce production-grade software, but they take months and require headcount most companies do not have. Pit’s claim is that AI now collapses the time and team size required to produce real production software, but only if the system around it handles the governance, observability, and infrastructure that enterprises actually need.
The product splits into two components. Pit Studio learns how a company works and builds the system that runs it. Pit Cloud is the governed infrastructure underneath, with tenant isolation, ISO 27001, SSO, RBAC, and full audit observability. The architecture is the unfair advantage. Without Pit Cloud, the output is just another vibe-coded prototype. With it, the output is production software.
The Pilot Numbers
Pit is already live with enterprise pilots in logistics, telecom, e-commerce, and healthcare. Named deployments include Voi, Tre, Stena Recycling, and Kry. The disclosed results are concrete. An 85% reduction in campaign execution time. More than 10,000 hours saved annually per deployment. A 99% invoice acceptance rate through automation. At one of Europe’s largest industrial companies, Pit replaced legacy contract and invoice validation with a real-time AI-powered system that has produced zero validation errors at deployment.
The TAM and the a16z Thesis
The market Pit is positioned inside is not small. The AI-native custom enterprise software segment is projected to grow at roughly 98% CAGR through 2030, the fastest-growing category within the broader AI software stack. Total AI-native enterprise software TAM is projected to reach approximately $530 billion by 2030, with custom-built software representing roughly half of that pool. The expansion is being driven by exactly the build vs buy migration the SaaSpocalypse priced in.
Alex Rampell at Andreessen Horowitz framed the investment in terms that read as a deliberate distancing from the broader AI hype cycle. “Every AI company is selling speed. Pit is selling speed that holds up for years, secure, governed, and built to last. It’s a new category,” Rampell said. The framing matters because the post-SaaSpocalypse market is now flooded with AI tools promising to do everything, and the category that survives the next eighteen months will be the one that pairs velocity with the boring requirements that enterprise procurement actually cares about.
What to Watch
Three signals will determine whether Pit defines the category or shares it with competitors.
The first is whether the pilot logos convert into multi-product enterprise standardization. A 10,000-hour saving on one workflow is a case study. A multi-year contract to run twelve workflows across an enterprise is a category.
The second is whether the AI product team as a service framing holds against competitive pressure from incumbents. Salesforce, Microsoft, and ServiceNow are not going to cede the operational layer without fighting for it, and the obvious counter-move is bundling AI app generation into existing platforms.
The third is whether the build vs buy curve continues bending toward custom. If Retool’s 35% becomes 50% by end of 2026, the entire SaaS layer becomes a transition asset and the operational layer becomes the contested ground. Pit is positioned for that scenario. The next four quarters will say whether the positioning was early or late.
The Klarna playbook worked inside one company and helped reshape an industry. Pit’s bet is that the same playbook, productized and accessible to every enterprise that cannot build it themselves, reshapes how all of them run.
Don’t forget to like and share the story!