Sionic AI brings its Thai-reading model to SEA, with banks as the real exam
Most AI models read Thai the way a tourist reads a Bangkok street menu: confidently, and often wrongly. Sionic AI says it has closed part of that language gap, and it is using the result to push into Southeast Asia. The Seoul-based enterprise AI company said last week it is expanding in Thailand and Singapore […] The post Sionic AI brings its Thai-reading model to SEA, with banks as the real exam…
Sionic AI, a Seoul-based enterprise AI company, has expanded its operations to Thailand and Singapore by leveraging its advanced Thai-reading model, PepperOCR-VL. This model has achieved top rankings in multilingual benchmarks, including MDPBench for Thai and Korean languages, and MTEB for Japanese and Korean language retrieval and reranking.
The success of PepperOCR-VL is attributed to its ability to run on a single graphics card and its availability on Hugging Face, making it accessible for local data centers. In Thailand, Sionic has partnered with Siam AI and local firm BlueZebra to develop an AI agent platform called Storm, a no-code tool allowing businesses to create AI assistants for multi-step tasks.
This partnership comes as Thailand becomes a thriving enterprise AI market, with support from global players like Huawei Cloud and OpenAI, and the implementation of privacy guidelines under its Personal Data Protection Act. In Singapore, Sionic is working on a proof of concept for a customer service chatbot with a fintech company specializing in insurance services.
The trial is nearing completion, and the success of this project will depend on the region's strict regulatory environment, particularly for financial institutions using third-party AI.
Written by urgent.news from e27's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
