Nairobi-based Lua AI has raised $5.8 million in seed funding to scale a platform where AI agents execute entire business processes, not as assistants to human workers, but as autonomous operators handling multi-step workflows end-to-end. Norrsken22 led the round, with participation from Y Combinator, Flourish Ventures, P1 Ventures, and Enza Capital.

From Assistance to Execution

The company’s core argument is that enterprise AI has spent five years optimizing within existing software boundaries (chatbots, copilots, summarizers) rather than removing those boundaries entirely. Lua’s platform lets organizations deploy agents that take a request, such as a loan application, and carry it through data collection, KYC verification, rule application, and escalation without handing off between systems or teams.

“We’re in the race to shape how human-agent collaboration gets defined globally,” co-founder Lorcan O’Cathain told TechCabal. “Organisations will be blends of humans and AI agents collaborating.”

The agents operate through channels businesses already use: WhatsApp, Slack, email, and voice. No new interface required.

Kenyan Financial Services as Proving Ground

Early deployments concentrate in Kenyan financial services, where manual processing delays are a measurable operational constraint. Retail loan approvals that require manual KYC checks and document verification routinely take multiple business days at traditional banks — a cycle Lua’s agents compress by automating the verification and approval steps end-to-end. Lua’s agents compress that cycle by automating the verification and approval steps, escalating to humans only when uncertainty exceeds confidence thresholds.

The model is already generating unconventional team structures. According to TechCabal, some early users are running teams of 1 to 3 people coordinating 10 or more agents, with some structuring agents as the core product itself rather than a support tool.

The Execution Threshold

Founded in 2023, Lua sits at the boundary between assistive AI and autonomous execution. O’Cathain argues the threshold shifted recently: “A year or two ago, AI could summarise documents and generate text, but it couldn’t reliably execute multi-step workflows with real business logic. That changed fast.”

PwC data cited by the company suggests 79% of US companies are actively exploring AI integration, though most remain in early deployment phases. Lua is betting that the next wave of adoption moves from augmentation (helping people do their jobs) to execution (agents doing the job).

Market Context

The funding positions Lua as one of the few Africa-based startups building for the autonomous agent layer rather than the copilot layer. While Microsoft, Google Cloud, and AWS dominate the assistive tooling market, and automation firms like UiPath extend existing workflows, Lua is targeting a different category: systems that complete structured business processes with minimal human input.

The company gives clients direct control over agent configuration and behavior, a design decision O’Cathain attributes to operating in markets where trust in software vendors is still evolving. Human staff remain responsible for oversight, exception handling, and final approval in sensitive workflows.

“One of the most valuable skills someone will have will be the ability to manage agents and help improve them,” O’Cathain told TechCabal.

Published: April 25, 2026