AethexAI, A Full-Stack Voice AI Infrastructure Startup, Raises $3M In Pre-Seed Round To Level The Playing Field For Enterprises In Emerging Markets – AfroTech

Mariama D. Diallo and Ayooluwa Odemuyiwa followed a shared conviction to leave their careers to launch a startup.

Diallo, who was born in West Africa and raised in Kentucky, worked in investment banking at Goldman Sachs for 4 years. He earned an MBA and a master’s in international and global studies at the University of Pennsylvania before briefly working at Centerview Partners in banking, then went on to work in product and growth at Model ML.

Odemuyiwa has lived in Canada, Alabama, and Missouri and earned a bachelor’s degree in computer science at the California Institute of Technology. She conducted research in computer vision and landed internships at Meta over three summers across different teams. She then took a full-time role as a software engineer at Meta for six months immediately after graduating, where she on Meta’s Instagram Reels and Oculus teams, before joining AeroVironment as a software engineer to work on mission-critical drone software.

Diallo and Odemuyiwa met through a mutual connection when Odemuyiwa was applying to the MBA program at the Stanford Graduate School of Business.

“We stayed in touch because we had a joint passion and excitement about the potential of tech in emerging markets from both of our backgrounds,” Odemuyiwa told AFROTECH™ .

Odemuyiwa worked on several voice AI projects and noticed that, while many Silicon Valley startups were building voice AI, few focused on the markets she had been researching. Meanwhile, Diallo observed a growing preference for voice-based AI interactions while speaking with bankers at Model ML.

The pair decided to start building a solution in July 2025, which led them to hop on a plane and travel to various countries, including the UAE, Egypt, Abidjan, and Lagos, to identify barriers for the technology.

They went to a call center in Egypt, which had automated 7% of its calls but rolled it back because it wasn’t efficient.

Odemuyiwa noted that the process at the call center was highly manual, with people listening to and responding to feedback on thousands of calls. The software used was outdated and “buggy,” which led to lost recordings and a lack of quality control.

The pair also went to the Ivory Coast and sat in a company’s call center, where they explored use cases for voice agents and whether the company was using local or working languages.

“In the Ivory Coast, the national language is French. We assumed that the calls that we’d be seeing would be done in local languages, so we provide more value by training our models to speak more indigenous languages,” Odemuyiwa said. “But we found that for a call — it was at 400,000 a week —we’re looking at maybe just three percent of those calls being in an indigenous language, whereas the rest of them were actually in French. That’s where we realized that there’s more emphasis on just making sure we understand how people speak the national languages from the beginning.”

AethexAI

Based on feedback from their travel experiences abroad, Diallo (chief executive officer) and Odemuyiwa (chief technology officer) built a prototype that has since become an official product. The pair founded AethexAI, a London-based full-stack voice AI and infrastructure startup targeting enterprises across emerging markets, including banks, telecom companies, call centers, and fintechs, with use cases spanning sales, outbound, loan recovery, and know-your-customer processes.

Additionally, it supports dialects of French and English and, later this year, will also support Arabic. The platform has more than 100 voices across French and English dialects, trained on hundreds of hours of source audio.

AethexAI’s first product is Kora 1, which is described as a “production-grade model stack trained on real conversational speech with human-labeled data across emerging markets,” according to its website.  Kora’s workflow engine can read and write data, perform actions, and connect to existing systems such as Gmail, WhatsApp, and call center software.

Beyond being a voice agent company, AethexAI can also handle backend workflows, the founders emphasized.

“We actually have the agents that do tool calling, which is when, during the call, it actually will go into your CRM and update it for you … If we’re doing an appointment and our agent picks up the call, if we’re already integrated, it’s able to book that appointment for you,” said Odemuyiwa.

AethexAI also offers simulation tools that create agentic personas to test its voice agents. This is helpful for evaluating performance across thousands of scenarios and optimizing prompts and workflows for use cases.

Additionally, AethexAI offers detailed analytics. According to the company’s website, every call is transcribed, scored, and fully searchable. Analytics include resolution rates, sentiment tracking, and agent performance metrics.

“How can we help companies create value the same way as companies in the West are using voice agents to be more efficient?” said Diallo. “We just want to level that playing field for enterprises.”

Funding

AethexAI has raised $3 million in pre-seed funding in a round led by 4DX Ventures, Enza Capital, Dorm Room Fund, Mojo Ventures, and Stanford GSB 26 Fund, with angel investors from Stanford faculty also participating, TechCrunch reports. The funding will be used to scale AethexAI’s team, improve its tech, and deploy more voice agents.