← All posts

Introducing Aiocia

December 8, 2024 ยท Aiocia Team

Artificial intelligence is moving fast. New models, new benchmarks, and new promises arrive every week. In the middle of all that noise, we asked ourselves a simple question: why not build something that actually helps people? That question became Aiocia.

We are a new AI company fueled by a "why not?" spirit. We approach every problem with curiosity rather than caution, cut through the hype to find what genuinely works, and keep our focus on one thing: practical AI that improves everyday life.

Our roots

Aiocia grew out of the instruct.kr open-source LLM community, where engineers and researchers gathered to share insights, run experiments together, and push the limits of what language models could do in Korean. That community became a proving ground for ideas that went on to matter well beyond the forum.

The most concrete example is LogicKor, a benchmark our founders developed within instruct.kr to evaluate LLM performance in Korean across a range of reasoning, comprehension, and instruction-following tasks. LogicKor earned wide respect as a standard for Korean LLM evaluation — used by researchers and companies alike to measure where their models truly stood. The discipline of building honest, reproducible evaluation is something we carry into every product we ship.

The team

Aiocia was founded in 2024 by three people with complementary backgrounds and a shared conviction that AI should be useful above all else. We operate as a remote, borderless company with team members across the United States, Canada, and South Korea.

  • maywell is an experienced LLM researcher who contributed to teams at Yanolja and Sionic AI before co-founding instruct.kr and creating LogicKor. At Aiocia, maywell focuses on delivering practical, impactful AI solutions rather than purely theoretical concepts — and on keeping the team honest about what models can and cannot do.
  • Sigrid Jin took a winding path from mathematics to blockchain data engineering at DSRV and machine learning engineering at Sionic AI. Sigrid brings deep expertise in Kubernetes, cloud-native infrastructure, and Python, and applies that knowledge to building systems that are reliable enough to put in front of real users.
  • Cilio comes from a product direction background, with experience managing content-driven projects and a sharp eye for media trends. Cilio shapes the LLM-enabled services Aiocia ships, steering product development toward experiences that genuinely resonate with audiences.

What we are building

Our flagship product is Elyn AI — a platform that reimagines how people interact with virtual characters. Character-based chat is a compelling use case for large language models: it demands consistency, memory, persona fidelity, and the ability to retrieve and reason over long conversational histories. Getting it right requires solving the same retrieval and generation problems that matter across the entire AI landscape, which is why we think of Elyn as both a product and a research environment.

We are building Elyn with a modular architecture. The components we develop — retrieval pipelines, embedding strategies, rerankers, evaluation tooling — are designed from the start to be reusable, so other developers and teams can integrate them into their own projects. We are also partnering with creators such as Teddy Note to bring compelling characters and storylines to the platform.

Beyond Elyn, our research interests span large language models, embedding and reranking techniques, and retrieval strategies. We believe that rigorous evaluation — the same instinct that produced LogicKor — is the foundation of trustworthy AI, and we apply that instinct at every layer of what we build.

Looking ahead

Aiocia is registered in South Korea and operates with a global outlook. Distributed teamwork is not a constraint for us — it is a feature. We believe that the best teams draw on diverse perspectives and time zones, and that the tools for remote, asynchronous collaboration have finally caught up with that belief.

Our vision is an AI ecosystem grounded in accessibility, reliability, and genuine utility: technology that adapts to human needs rather than demanding that humans adapt to technology. That vision starts with the products we ship today and extends to the tools and research we contribute to the broader community.

We are just getting started. If you are curious about what we are building, want to explore a partnership, or simply have a question, we would love to hear from you. Reach out at [email protected].