Tech

AethexAI secures $3 million pre-seed to build voice AI infrastructure for Africa and Middle East

With 4DX Ventures leading the round, AethexAI is deploying its proprietary Kora series models to handle over 17,000 daily calls, targeting debt collection, customer activation, and KYC verification.

Author
Owen Mercer
Markets and Finance Editor
Published
Draft
Source: TechCrunch · original
These two founders left Goldman and Meta to build voice AI for markets everyone else overlooked
Startup founded by ex-Goldman Sachs and Meta executives develops custom small language models to address latency and dialect challenges in emerging markets

AethexAI, a voice artificial intelligence startup founded by former Goldman Sachs and Meta employees Mariama Diallo and Ayooluwa Odemuyiwa, has raised $3 million in pre-seed funding. The investment was led by 4DX Ventures, with participation from Enza Capital, Dorm Room Fund, Mojo Ventures, and the Stanford GSB 26 Fund. Individual investors include Stanford faculty, telecom executives, and AI researchers from Anthropic. The capital will support the development of voice AI infrastructure specifically tailored for Africa and the Middle East, markets that have historically been overlooked by major global players.

The company has developed its own stack rather than relying on existing orchestration tools such as Vapi or LiveKit. This decision was driven by the need to reduce latency and accurately handle local dialects of English, French, and Arabic. AethexAI built its own orchestration layer and developed the Kora series of small models, ranging from 300 million to 1.7 billion parameters. These models are significantly smaller than large language models, a design choice intended to maintain accuracy while minimising the delay that often plagues automated calls in regions with less robust digital infrastructure.

Data acquisition and training were conducted through unconventional means to keep costs low and ensure local relevance. The startup utilised anonymised recordings from a call centre partner and shipped hard drives to radio stations across Africa to gather audio data. To assist with annotation and pronunciation of local names, AethexAI established a contributor network of university students. This approach allowed the company to train its models effectively without the massive expenditures typical of AI labs deploying large-scale systems.

Currently, AethexAI is processing more than 17,000 calls per day. The platform is launching its enterprise suite, including APIs and SDKs, allowing businesses to integrate the technology. Key use cases identified by the founders include debt collection, customer activation, and Know Your Customer (KYC) verification. The company is hiring forward-deployed engineers on a contract basis and building channel partnerships with telecom providers to manage telephony for these voice AI interactions.

Walter Badoo, co-founder and managing partner of 4DX Ventures, noted that enterprises in Africa and the Middle East process roughly three times the call volume of their Western counterparts. He argued that incumbent systems were built for Western markets characterised by high-end GPU infrastructure and standard English speech environments, creating gaps when applied to regions with diverse dialects and code-switching. AethexAI aims to fill this void by providing infrastructure that works within existing local telephony systems and price points, betting that the gaps left by global giants represent a significant market opening.

Continue reading

More from Tech

Read next: Apple to roll out manual EQ controls for AirPods in iOS 27 update
Read next: Apple rolls out visionOS 27, integrating AI-driven Siri into Vision Pro headset
Read next: Apple Overhauls Siri with Google Gemini Partnership and Standalone App at WWDC 2026