Editor's note: This article is part of [Swipe], a series in which we open up and examine AI apps and services making waves in South Korea. This edition covers Daglo, a voice-recording service by ActionPower.
When a meeting ends, what remains is a recording file and a sinking feeling. Replaying an hour-long interview and typing it up by hand is a task AI now dispatches in roughly a minute. In a comparative test conducted by Digital Daily in July of this year, both Daglo and Clova Note — Naver's rival service — transcribed a 67-minute interview in under a minute each. Transcription accuracy, in other words, is no longer a point of differentiation. So what, then, is Daglo competing on?
What is Daglo?
Daglo is a voice-recording service operated by ActionPower, an AI start-up. According to Digital Daily, the company launched with voice dictation in 2017. Its official website now describes the product as a "voice intelligence workspace" that captures speech from meetings, lectures, videos, and uploaded files, then uses AI to transform it into whatever output the user needs. The homepage states that the platform has surpassed 2m cumulative users and transcribes more than 13m hours of audio per year — figures provided by the company itself. An earlier report by MoneyToday in March 2025 cited more than 1.5m users and 10m hours of transcription in the previous year.
Pricing is divided into a free tier and paid plans — Pro, Premium, and a team subscription. Under the free plan, users receive 1,000 credits per month; paid plans offer more credits and additional features.
Core features
The basics are recording, file transcription, speaker identification, and meeting summaries. On top of these, Daglo adds YouTube-link summarisation, PDF analysis, AI-assisted document translation, and the automatic generation of slides and quizzes.
The most distinctive feature is "Board Chat." Rather than re-uploading materials, users can query the contents of previously saved recordings and documents, request summaries, or commission written outputs — all within a single interface. Digital Daily noted that Board Chat groups results by category and attaches timestamps, making it easier to cross-reference the original source.
Daglo also lets users choose from multiple large language models within a single platform. During a major upgrade in March 2025, ActionPower added the ability to select from seven leading LLMs — including OpenAI's GPT series — on the desktop version, along with 22 document templates. The upgrade was marketed under the slogan "All AI, in Daglo, all at once." ActionPower also runs its own summarisation-optimised model called "Ellie." Lee Ji-hwa, co-chief executive and CTO, has stated that the company's proprietary speech-recognition engine achieves roughly 95% accuracy — again, a figure from the company itself.
Strengths and limitations
Daglo's chief strength is that it handles the work that comes *after* transcription within the same screen. In Digital Daily's comparative test, Daglo proved more accurate at recognising proper nouns and personal names; Clova Note, by contrast, incorrectly divided a two-person conversation into three to six speakers. On hallucinations, Lee Ji-hwa explained that "Board Chat generates answers based on uploaded materials, so it produces fewer hallucinations than general generative AI." In PDF summarisation, Daglo flagged when a concept did not appear in the source document and cited page numbers.
There are, however, three notable limitations.
First, transcription errors carry over into summaries uncorrected. In the same test, the term "human-in-the-loop" was rendered as "humanism" in the transcript — and the summary did not catch the mistake. Template-based summarisation also produced a hallucinated company name that did not appear in the source text.
Second, information density. When summarising a PDF, ChatGPT generated ten bullet points; Daglo produced six.
Third, and most seriously, security. On 22nd March this year, an external ransomware attack brought the service down. According to Byline Network, both the app and web servers were taken offline before full functionality was restored by 8am the following morning. A company representative apologised, describing it as "a temporary service disruption." In its Digital Daily coverage, Daglo stated that an investigation conducted alongside the Korea Internet & Security Agency (KISA) found no evidence that users' personal information or voice data had been leaked — though it acknowledged that the possibility of further damage could not be entirely ruled out. The company pledged to obtain ISO 27001 certification and establish a continuous monitoring system to prevent a recurrence. For a service to which users entrust sensitive recordings, this episode is not easy to brush aside.
Assessment
Daglo is in the process of evolving from a transcription tool into something closer to a "records-to-results" platform. According to Digital Daily, its near-term roadmap includes integration with calendar apps and with Slack and Notion, as well as shared team boards. Greater connectivity will broaden the product's utility — but as document-specialised AI proliferates, Daglo will need to keep demonstrating why it deserves a place in that increasingly crowded landscape. For journalists and office workers, it offers genuine time savings; but summaries should be cross-checked against originals rather than taken at face value.
★★★★☆ (3.5/5.0)
Verdict
"Transcription is already a commodity. Daglo's real test lies in the connectivity that turns records into work — and in the trustworthiness required to keep sensitive data safe."