He Was Four Weeks From Law School. He Chose Engineering Instead.

Four weeks before law school started, Nizar Abi Zaher changed his mind.

He had spent the preceding months teaching himself to code — not as a hedge, but because the pull toward building had become something he could not argue with. Law was the plan. Engineering was the pull. He chose the pull.

His co-founder Adam Wazzan had never faced that same conflict. Adam grew up in Syria, moved to Canada, and had always been a builder in the literal sense — hardware, software, anything that required making something that did not previously exist. When Nizar and Adam met, the orientation was the same: figure out what is missing, then build it.

What was missing, it turned out, was nine months away.

Orchid, and the nine months that became Keiki

Nizar and Adam spent nine months building Orchid, a personal AI assistant that became a Y Combinator Spring 2025 company. The goal was to make AI feel effortless — to take the models, the APIs, and the capabilities that already existed and turn them into something a person could actually rely on.

Making something feel effortless, it turns out, requires an offensive amount of effort.

The AI models were already exceptional. The APIs were accessible. What required nine months was everything between a working demo and a product that works reliably: messaging infrastructure, memory management, tool integration, hosting, and the specific ways things fail when real users arrive and expect real reliability.

"Nobody starts a company because they're passionate about debugging webhooks at 2 a.m.," Nizar says.

That line carries everything. It is not a complaint — it is a diagnosis. The gap between what AI makes possible and what shipping actually requires is real, expensive, and almost entirely invisible from the outside. Every team that wants to build something on AI is about to pay some version of that tax.

When Orchid launched, something unexpected happened. Companies started reaching out asking how Nizar and Adam had built such a capable, reliable iMessage agent. They wanted that foundation for their own products.

Nizar and Adam realized they'd spent nine months building infrastructure that other teams were about to spend nine months rebuilding.

So they made it available. That became Keiki.

What Keiki does

Keiki lets companies build and launch AI agents on iMessage, WhatsApp, Slack, SMS, Telegram, and email — deploying to the channels their customers already use.

The problem it solves is not the AI itself — models are abundant and improving. The problem is the infrastructure layer between a great idea and a product customers can actually use: the messaging, the memory, the tools, the hosting, and the failure modes that only appear in production.

Keiki handles that layer. The founder brings the product and the ambition.

The opportunity Nizar sees is straightforward: companies spend enormous amounts of money getting customers to a website or into an app. Messaging gives them another channel — a conversation that customers can return to whenever they need something. A customer can ask which product fits their needs, get a recommendation, and complete a purchase. They can use a service by sending a message. They can pick up exactly where they left off.

"Your next purchase might start with 'hey,'" Nizar says.

That is the version of commerce and service Keiki is built to power.

The team ratio that is not an accident

Keiki shipped on September 1, 2026 with five people: four focused on growth, one on engineering. In an AI company, that ratio looks like a misprint. It is not.

Nizar's reasoning is explicit: AI is making coding less of a bottleneck. The functions that previously required a large engineering team can increasingly be handled with fewer people and more capable tools. What is not becoming less scarce is the ability to understand customers, communicate clearly, and get people to genuinely care about what you are building.

"Understanding customers, communicating clearly, and getting people to care are becoming even more valuable," Nizar says. "Our team reflects that belief."

This is a bet — a deliberate one, made with evidence from how Orchid was built and what Keiki is being built for. The infrastructure is handled. The question the rest of the team is trying to answer is: how do you reach the people who need this, and how do you make them understand what it does for them?

The ambition behind the business

Nizar and Adam are explicit about what they are trying to build toward.

WordPress made publishing accessible to people who could not build a content management system from scratch. Shopify made launching an online store accessible to people who could not build ecommerce infrastructure. Keiki wants to do for AI agents what those platforms did for their categories.

"We believe every business will have an agent, just as businesses came to have websites," Nizar says. "People will reach those agents to ask questions, buy things, and get work done."

The version of that world Keiki is positioning toward is one where a small team can offer a useful, personal, AI-powered experience to customers anywhere in the world — without first spending nine months building the infrastructure that makes it possible.

That is what Orchid cost. That is what Keiki is designed to give back.

---

Nizar Abi Zaher is the CEO and co-founder of Keiki, with Adam Wazzan as CTO and co-founder. Keiki lets companies build and launch AI agents on iMessage, WhatsApp, Slack, SMS, Telegram, and email. The platform is available at onkeiki.com.