AethexAI raises $3 million to power localized AI calls in Africa and Middle East

Silicon Valley spent years treating voice AI as a problem of model size and developer tooling. AethexAI is betting the bigger blind spot was geography. The startup said its stack is now handling more than 17,000 calls per day across Africa and the Middle East, a level of usage that helped justify a $3 million pre-seed round led by 4DX Ventures, with participation from Enza Capital, Dorm Room Fund, Mojo Ventures and Stanford GSB 26 Fund.

Founded last year by Mariama Diallo and Ayooluwa Odemuyiwa, AethexAI was built around a simple argument: in markets where English, French and Arabic are spoken in local dialects, the usual playbook of plugging into off-the-shelf orchestration tools was not enough. Instead of relying on systems such as Vapi or LiveKit, the company built its own small model and orchestration layer from scratch, then tuned its Kora model family to sizes ranging from 300 million to 1.7 billion parameters. The goal was to cut latency at every step, because Diallo and Odemuyiwa said automated calls in the region were plagued by jitter and delays that made larger, externally hosted models impractical.

The founders’ thesis came out of direct market pain. In Egypt, they found a call center that had automated a significant share of its calls before rolling the system back after poor results. They also heard the same complaint from support centers across Africa: hiring engineers to automate voice operations at an affordable cost remained difficult. That reality has broader implications for the AI industry. Markets with less predictable connectivity, mixed-language customer bases and tighter operating budgets often need smaller, faster models rather than the largest frontier systems. For enterprises, that can mean the difference between a product demo and a working deployment.

Diallo previously worked at Goldman Sachs and later at YC-backed ModelML, where a speaker bio says she was also a product and growth hire and the first go-to-market employee. Another bio says she also worked at Centerview Partners and earned an MBA from Wharton and a master’s from the Lauder Institute at the University of Pennsylvania. Odemuyiwa graduated from Caltech, worked at Meta and enrolled at Stanford Business School before co-founding AethexAI; a Caltech profile says she also became an outspoken advocate for student mental health.

AethexAI is now opening its platform to enterprises and giving developers APIs and SDKs to test its models. The funding round, backed by investors from Brooklyn to Stanford, signals a widening bet that the next major wave of AI infrastructure will not come only from the usual U.S. enterprise markets, but from regions that were long overlooked and are now demanding software built for their own languages, latencies and economics.