Habari, Kenya.

I'm Matt, founder of AirDial. As of today the product is live in Kenya.

Before I tell you what we built, an observation about this market I think is worth your time.

The customer expectation Kenya already set

M-Pesa now processes 100 million transactions a day in Kenya. 40 million monthly active customers. KSh 38.29 trillion in annual transaction value — roughly $292 billion. The Fintech 2.0 platform Safaricom rolled out in September 2025 lifted the per-second capacity to 6,000 with a ceiling of 12,000 at peak, and at Safaricom Decode 4.0 in April 2026, the company formally declared 2026 its "Year of AI." Kenyan financial inclusion is now 84.8%, up from 19% before M-Pesa launched.

What this produces for every other Kenyan business is a customer who has been spoiled by good infrastructure for almost two decades. They expect a transaction to take three seconds. They expect a callback in minutes, not days. They expect to be reached on the channel they actually use — overwhelmingly WhatsApp for messaging, voice for anything that needs a human relationship — in the language they actually use. That mix of English, Swahili, and Sheng is not a localisation problem. It's the substrate.

The gap between the response speed Kenyan customers expect and the response speed traditional contact centres can deliver is not closeable with more headcount. Peak in this market means simultaneous, not sequential — M-Pesa trained the country to act at the same time. You cannot hire fifty agents in Nairobi and have them keep up during peak.

The lesson from the Safaricom AI lawsuit

In April 2026 a constitutional case was filed against Safaricom over its use of AI in customer service — the petition argues, in part, that the Zuri chatbot acts as a gatekeeper between customers and human agents in cases where human intervention is genuinely needed. This is worth taking seriously. The risk for any business deploying AI voice or WhatsApp in Kenya right now is that the technology gets built as a wall between the customer and the company, rather than as an extension of the company's ability to actually serve them.

The right design is the opposite. AI handles the high-volume repetitive work that should never have required a human in the first place. Confidence scoring decides when to keep going and when to hand over. Escalation to a human is a first-class path, not a fallback hidden behind ten menu options. The audit trail is exportable. The customer always knows they can speak to a person if they ask. This is the design our product enforces by default.

What AirDial is

AirDial is the AI voice and WhatsApp infrastructure for Kenyan businesses. Local +254 numbers across Nairobi, Mombasa, Kisumu, and the rest of the country. Voice and WhatsApp agents that handle English, Swahili, and Sheng natively. M-Pesa awareness is built in at the conversation layer — an agent can reference, prompt, or confirm an M-Pesa transaction inside the same call it's having about something else. Sub-200ms voice latency on Kenyan routes. The Data Protection Act 2019 is built into the architecture — consent capture, data residency, recording controls, audit logs. Pricing visible on the site, in shillings or USD.

What you'd actually do

Sign up at airdial.co, build an agent around the workflow you actually want to scale — collections, lending follow-ups, appointment confirmations, support escalations, lead qualification — and watch it run. Pricing visible from the first page. No salesperson.

The Kenyan operators who deploy AI voice in 2026 with the right design — automation where it adds value, graceful escalation where it doesn't — are the ones who will set the standard for what good looks like in this market for the rest of the decade.

It's nicer over here.