This article is part of "Swipe," a series examining Korean AI apps and services encountered in everyday life. This instalment looks at QANDA, the AI-powered learning app developed by Mathpresso.

Introduction

QANDA is a learning app that made its name with a disarmingly simple proposition: photograph a problem you cannot solve and receive a worked solution in return. Its developer, Mathpresso, claims on its official website that the platform has accumulated more than 100m registered users, more than 5m monthly active users, and processes an average of 3m questions per day. The company also states that half of all South Korean secondary-school students use the app. These figures are self-reported and have not been independently verified.

One detail stands out: the geographic composition of its user base. When the company announced that registrations had surpassed 70m in October 2022, it disclosed that 87% of those users were based outside South Korea. Vietnam accounted for 27% and Indonesia for 24%. Though QANDA is widely perceived as a domestic Korean student tool, it has in fact grown its audience predominantly in South-East Asia.

Core Features

The centrepiece is AI-powered problem-solving. When a student photographs a question, optical character recognition (OCR) technology reads the text—including mathematical notation—and generates a solution. According to details released in February 2025 when Mathpresso launched a co-branded subscription product with South Korean telecoms operator KT, the process takes roughly five seconds. The output includes step-by-step workings, hints, and explanations tailored to the individual student, along with follow-up prompts designed to encourage independent thinking.

Students can also photograph their own attempted solutions and receive annotated feedback—identifying not merely whether the answer is wrong, but where the reasoning went astray. Alongside AI-generated content, the platform hosts video and handwritten solutions from what the company describes as more than 100 vetted top-ranked local tutors and teachers from leading universities. A direct Q&A channel connects students with human instructors.

Rounding out the offering are flashcard-style quizzes for memorising concepts and a study-time timer that logs focused hours—standard features for an app positioning itself as a comprehensive study companion.

Mathpresso has also experimented with an AI tutoring persona. In August 2023, it launched a beta version of "Polly," an evolution of "QANDAbot," a large-language-model-based solution assistant it had introduced in March of the same year. Polly was described as placing greater emphasis on dialogue—asking targeted questions to guide students towards answers rather than simply delivering them. How Polly currently features within the app could not be confirmed.

Pricing follows a freemium model. Free users receive a set number of daily questions at no charge. A premium subscription costs ₩18,500 per month (roughly $14), unlocking 30 features without restriction and removing advertisements. KT subscribers can access an ad-supported "Premium Lite" tier for ₩11,000 per month.

Strengths and Limitations

QANDA's principal strengths are speed and breadth. A single photograph yields solutions spanning all subjects, and a two-track structure—AI first, human teacher second—means students who cannot follow the automated explanation have somewhere to turn. The design reflects an ambition to shift the emphasis from answer retrieval towards genuine comprehension. Mathpresso has articulated this position consistently: in a 2021 interview with AI Times, the company described QANDA as "a learning tool that helps students study" and claimed that students using it merely to copy answers could be identified by behavioural patterns—flicking past solutions without reading them, or searching through a string of different problems within seconds.

The business is growing. According to MoneyToday, Mathpresso's consolidated revenue for 2024 reached ₩24.2bn, up 42% year on year. The operating loss narrowed by around 20% to ₩19.4bn, from ₩24.2bn the previous year. The company remains loss-making despite its scale, and the freemium architecture—daily free questions designed to nudge users towards a paid subscription—appears central to its commercial strategy.

Three limitations are worth noting. First, the risk that the app substitutes for homework rather than supporting learning. In 2021, a Mathpresso representative said internal analysis of usage patterns suggested this risk was not material, and that no specific countermeasures had been introduced. Whether that assessment has been revisited since is unclear. Second, accuracy. The platform's marketing emphasises "precise answers," but no independent verification of the AI's error rate could be found. On the American App Store, QANDA holds a rating of 3.5 out of 5, based on 772 reviews—too small a sample to be meaningfully representative. Third, price. At ₩18,500 a month, the premium tier is a non-trivial commitment for a student.

Verdict

QANDA is the longest-standing AI problem-solving app in the South Korean student market. It has evolved from a single elegant trick—photograph a question, receive an answer—into a broader ecosystem encompassing feedback, teacher access, and revision tools. The fundamental question, however, endures. Whether seeing a worked solution translates into genuine understanding depends less on the app than on the student using it. QANDA's investment in step-by-step explanations and follow-up questions suggests the company is aware of this gap and is attempting to close it. For students who simply want an answer quickly, it is a convenient instrument. For those willing to follow the reasoning, it can serve as a useful guide.

★★★☆☆ (3.5/5.0)

In a word: the answer arrives in five seconds—what you do with it is up to you.