level: business
aethexai, a startup founded by former goldman sachs and meta employees, raised $3 million in pre-seed funding to build voice ai for africa and the middle east. the round was led by 4dx ventures, with participation from enza capital, dorm room fund, mojo ventures, and stanford gsb 26 fund. individual investors include stanford faculty, telecom executives, and ai researchers from anthropic. the company built its own small models and orchestration layer instead of using existing tools like vapi or livekit, aiming to handle localized dialects of english, french, and arabic.
the founders found that call centers in egypt and africa struggled with high latency and poor results when using large ai models hosted outside the region. to solve this, aethexai developed the kora series of models with 300 million to 1.7 billion parameters, much smaller than typical large language models. they trained these models using anonymized call recordings and audio data collected from radio stations across africa. a network of university students helped annotate data and pronounce local names, keeping costs low. the startup now handles over 17,000 calls per day.
on the business side, aethexai offers onsite demos and workshops to help clients new to voice ai pick one key use case to start with, such as debt collection, customer activation, or know your customer verification. the company hires forward-deployed engineers on contract and partners with telecom providers to manage telephony. investor walter baddoo of 4dx ventures noted that enterprises in africa and the middle east process about three times the call volume of western counterparts, with voice as the dominant channel. existing systems often fail due to dialect, code-switching, and infrastructure gaps, creating an opening for specialized startups.
why it matters: voice ai that works with local dialects and low latency can make automation practical in regions where voice is the main customer channel, reducing costs for businesses.