South Korean company Weekley has completed a feature upgrade to its local AI platform Owllo (아울로), adding AI intelligent agents, skill plugins, natural-language photo search, and other capabilities that dig deep into local documents and photo data on a PC. At the same time, the company is extending into hardware, preparing a US launch of a 10-inch color ePaper device called Paper, with plans to connect the hardware with the Owllo platform.

Owllo is an open-weight large-model local runtime platform designed for ordinary users. Without manually downloading models, configuring a terminal environment, or entering API keys, users can directly invoke multiple local large models on their personal computers. Compared with early test versions, the new Owllo distinguishes between ordinary chat mode and intelligent agent execution mode, adds independent task and material management sections, and offers extensible skill plugins such as email handling, schedule management, file organization, and web search — lowering the barrier for ordinary users to get started with local large models, with local task processing incurring no extra billing.

The platform focuses on strengthening the use of local personal data. It adds natural-language photo search, so even if users cannot remember a file name or shooting date, they can describe it in natural language — for example, "photos taken on a sunny trip last year" — and directly search image files on their computer.

The team believes this is precisely the core value of local AI. Using cloud AI such as ChatGPT or Claude to process private documents and photo libraries requires uploading data to external servers; uploading tens of gigabytes of photo material is not only cumbersome but also raises users' concerns about the leakage of sensitive private information.

Owllo adopts a hybrid local + cloud scheduling strategy: simple tasks are handled by lightweight local models; complex needs invoke higher-performance local large models on the machine; and for scenarios beyond the capability of local models, external cloud AI is called via API as a supplement after obtaining user permission.

Weekley CTO Seo Woo-seok said: "Repeatedly uploading tens of gigabytes of personal data to the cloud is troublesome, and many users are reluctant to hand sensitive materials to third-party service providers. But local models cannot yet fully replace top frontier large models, so local AI and cloud AI should be viewed as complementary rather than mutually replacing."

Beyond software iterations, Weekley is formally expanding into hardware. The 10-inch color ePaper display Paper is being prepared for sale in the US market. The product is positioned as a home auxiliary secondary screen that can display photos, schedules, weather, and other life information. In the future, personalized content such as schedules and images processed by the Owllo software can be pushed directly to the Paper ePaper screen for display.

Weekley CEO Yoon Ju-hyun, speaking about the product philosophy, said: "No matter how AI computes, the final output should be visible and perceptible to users. The Paper hardware exists precisely to integrate AI-generated content into daily life scenarios."