There is a particular kind of startup story that plays out quietly in Nairobi’s fintech circles, builds a product most people have not yet used, and then resurfaces a few years later with investors queuing up to write cheques. Lua, the AI agent company founded in 2024, just wrote one of those chapters.
The company has closed a $5.8 million (roughly €4.9 million) seed round led by pan-African growth fund Norrsken22. Participation came from Flourish Ventures, 20VC, P1 Ventures, Phosphor Capital, and Y Combinator. Angel cheques came in from Privy CEO Henri Stern, Opendoor CEO Kaz Nejatian, and Nuitee CEO Med Benmansour.
Although Lua is technically headquartered in London, its operational DNA is East African. CEO Lorcan O’Cathain spent years in Kenya, serving as COO at microlender 4G Capital, running the Africa business of Zephyr Management, and co-founding Money254, the financial products comparison platform with several million users in Kenya. His co-founder and CTO Stefan Kruger was VP of Engineering at Paystack before Stripe acquired it. The two met while scaling a fintech business in East Africa.
What Lua actually does
Drop the mental image of a chatbot. Lua is building what it calls an “operating system” for AI agents that work alongside human teams. In practice, a bank, lender, or insurer can use Lua to stand up AI workers that handle customer onboarding, loan processing, insurance workflows, and other high-volume tasks. When a task needs human judgement, the agent hands it off. When the human finishes, the agent picks up the next step.
The product has two surfaces. Developers can ship through a command line interface. Non-technical staff can use a visual interface, or build through natural language. The platform handles the usual plumbing underneath, including model orchestration, integrations, and monitoring, so customers focus on defining business logic.
Lua is deliberately rejecting the “pay more when your agents succeed” pricing that several competing agent platforms have adopted. “Most agent platforms compound this with black box tooling and per-outcome pricing,” O’Cathain said in the announcement. “Lua is built on the opposite principle: teams own their agents, own their outcomes, and build compounding efficiency over time.”
The traction
Since launching in October 2025, Lua reports revenue has grown close to 30% week on week. In February 2026 alone, more AI agents were built on its platform than in all previous months combined. According to outlet Inforcapital, Lua crossed $1 million in annual recurring revenue within its first three months.
One note of caution: week on week percentages off a small base can flatter the underlying business. The $1M ARR milestone is the more grounded data point.
Early customers read like a map of East African fintech ambition. Turaco, the Nairobi-based microinsurer that has insured more than one million Kenyans through an M-KOPA partnership alone, is running on Lua. So is Tushop (social commerce), Umba (digital banking), and Kampala-based Numida, which lends to Ugandan MSMEs. These are exactly the kinds of businesses that process millions of customer interactions a month and spend meaningfully on human agents.
Why Norrsken22 matters
Norrsken22 is worth a moment. Backed by Klarna co-founder Niklas Adalberth and 33 unicorn founders, the $205 million fund closed its debut vehicle in late 2023 and has since written cheques into African-rooted startups with global ambitions. Lexi Novitske, the fund’s Lagos-based General Partner, said Lua’s footprint across Africa, Asia, the US, and Europe gives it “a pricing intuition that’s difficult to replicate.
The capital will go to two places. First, expanding the developer community. Second, growing the Lua Implementation Network, a group of partners who deploy the platform for clients in their home markets. Planned expansion covers Kenya, the US, Europe, and Asia.
The bigger picture for Kenya
Lua joins a small but growing set of companies building AI agents for African workflows. Nairobi-based Phindor, the company behind JuaFlow, is tackling a similar problem from a WhatsApp-first angle. The two are not head-to-head competitors. JuaFlow leans toward retail and SME automation with strong local language support. Lua is aiming higher up the stack, at mid-market and enterprise financial services teams worldwide.
For Kenyan banks, insurers, and digital lenders, the practical question is no longer whether to deploy AI agents. It is which platform to build that workforce on, how much of the cost structure to hand over to a vendor, and how much to retain in-house. Lua is betting that the winners will own their stack rather than lease outcomes.
