Sierra

Agent Engineers and the Future of Forward Deployed Engineering

Natalie Meurer, Head of Agent Engineering at Sierra, discusses the evolving role of forward deployed engineering (FDE) in a Q&A published by Latent Space. She describes 'agent engineers' as customer-facing engineers who combine systems integration, orchestration of multiple models, and product/UX sensibility to build conversational AI agents for enterprise customer service. Meurer says the role is defined more by accountability to customers than a fixed skill set, that agent work often focuses on orchestration rather than changing underlying models, and that product and customer-facing engineering are beginning to converge. Sierra builds voice, chat and email agents and reports some enterprise customers reach production in 40–60 days.

AIEWF Day 2: Loops, Software Factories, Forward Deployed Engineers

Day 2 of the AI Engineer World’s Fair centered on “loops” and the emerging concept of software factories: iterative agent-driven workflows that automate the full software lifecycle. Speakers included AIEWF cofounder swyx, Microsoft’s Pablo Castro (on Foundry), and OpenAI presenters Alexander Embiricos and Romain Huet (on Codex and multi-agent loops). Presentations from Tereza Tížková (Factory) and Zach Lloyd (Warp) advocated building platforms that orchestrate agents across repo, feedback, logs and review cycles. The conference highlighted Forward Deployed Engineers (FDEs) as a new role for integrating and orchestrating agents in customer environments (notably Natalie Meurer at Sierra and Pauline Brunet at Cursor). The day also emphasized open-source model momentum: Z.ai’s Zixuan Li presented GLM-5.2 and ZCode; MiniMax’s Olive Song discussed M3; Ahmad Osman (Osmantic) spoke on local AI deployments.

AethexAI raises $3M to build voice AI for Africa, MENA

AethexAI, a startup founded in 2025 to build voice AI tailored to Africa and the Middle East, raised $3 million in pre-seed funding led by 4DX Ventures with participation from Enza Capital, Dorm Room Fund, Mojo Ventures and Stanford GSB 26 Fund. Founded by CEO Mariama Diallo (ex-Goldman/ModelML) and CTO Ayooluwa Odemuyiwa (ex-Meta/Caltech), the company built its own small models and orchestration layer — the Kora series (300M–1.7B parameters) — to reduce latency and handle localized dialects of English, French and Arabic. AethexAI used anonymized call-center recordings, radio-collected audio and a student annotator network to train models, says it now handles 17,000+ calls per day, and is launching an enterprise platform with APIs and SDKs focused on use cases like debt collection, customer activation and KYC.