Aotter

Before incorporation: PasteWall and flyingV, 2007–2012

Lun Hsieh (謝綸, also known as Konois) and Peng-han Su (蘇芃翰, “Dr. Porkchop”) belonged to Taiwan’s web-making and bulletin-board culture. They met through a school-age web competition.⁴ Hsieh studied library and information science at National Taiwan University; his formation combined information organization with practical website work. An NTU departmental account identifies him as an alumnus, while his professional biography describes freelance design work and experience as a technology editor and online writer.^{5, 6}

Su’s academic route was biological rather than a conventional software-engineering apprenticeship. A 2022 exhibitor profile identifies his doctorate in NTU’s molecular and cellular biology program. His subsequent work with retired biology professor Chen Chun-hung on educational games shows that this connection survived the software business: in the separate venture 變形漁房, Hsieh worked on brand strategy and Su on game planning. That venture should not be silently counted as an Aotter subsidiary.^{7, 8}

In the founders’ 2013 account, Su proposed PasteWall in 2007, influenced by the Mac and Microsoft’s tabletop Surface interface. Hsieh recognized a resemblance to something he had imagined on his own PTT2 page. Kao Chen-yang (高禎陽), later associated with Ontoo, encouraged development. The pair initially lacked the implementation skills they needed; a future technical chief joined in 2009, and a six-person core had formed by 2010.⁹

The present company identifies a third cofounder, Tzu-lin Huang (黃慈霖), an NTU information-management graduate who helped establish its early technical foundations and remains a technology adviser. Hsieh is CEO and company representative; Su leads research and development. These are complementary roles, not evidence that either public spokesman single-handedly made the organization.¹⁰

A retrospective chronology places the team’s formation of Ontoo in 2011 and its principal flyingV work between January and October 2012. Crowdfunding made sense alongside PasteWall: ideas developed collectively could eventually seek the money to be realized. But participation in the early flyingV team and ownership of today’s flyingV are different propositions. The team left that organization and incorporated Aotter on October 18, 2012.¹¹

Hsieh later attributed the separation to disagreements with investors and insufficient understanding of contractual relationships. This is his retrospective explanation, not an adjudicated account of fault. The important institutional consequence was a restart under a new company, with a continuing development team but without a ready-made financial base.¹²

The 2016 interview describes NT$300,000 assembled at the beginning and difficulty meeting the first month’s rent and payroll. It also records seventeen full-time and twenty-four part-time workers at that later date. Neither figure describes the current workforce. The distinction matters: the early institution was a mixture of continuing collaborators, employees and intermittent labor, rather than a fully staffed product company waiting for a launch.¹³

PasteWall, reusable software and 鄉民晚報

Contract work financed development. In November 2013 the founders said they had reduced commissions since July to concentrate on PasteWall. Its interface had simplified from a three-dimensional conception to one scrolling plane, without lines connecting notes. Approximately six hundred students in selected courses at National Taipei University of Technology and National Chengchi University supplied feedback. Public launch was still prospective.¹⁴

Aotter’s anniversary chronology dates the beta to 2014. It separately dates the first AotterClam backend to 2012 and the beginning of Trek development to 2013. The names therefore predate their current products: the original Clam was reusable backend machinery, not the enterprise AI service introduced fourteen years later. In 2018, CNA still described PasteWall as under development. “Beta released” and “finished, commercially established platform” cannot be used interchangeably.^{15, 16}

The more immediate breakthrough was 鄉民晚報, literally “Villagers’ Evening News,” with “villagers” referring to Taiwan’s online-forum public. In early 2013 Hsieh sketched its interface and Su used Clam to make a prototype. Initially, people selected links manually. As use grew, the team introduced a system that measured the rate at which likes and shares accumulated. Readers were sent to original publishers rather than kept inside a replacement copy of each article.¹⁷

Human editing remained substantial. In 2015 deputy editor Hsieh Wei (謝維) described reviewing roughly a thousand leading items daily, excluding conspicuously false or sensational material and adding short, colloquial comments. Reader objections prompted experiments with removing commentary or substituting excerpts; the distinctive voice returned. The Sunflower Movement had also exposed how a popularity-driven feed could lean heavily toward a particular political public. Its output combined a mathematical selection rule, the behavior of participating communities and editorial decisions.¹⁸

A Taipei government profile credits referral traffic with attracting media cooperation and providing early commercial income. This establishes a plausible route from usefulness to business: publishers gained mobile readers, while Aotter gained integrations and relationships. It does not identify the first paying publisher, disclose its invoice or establish which contract first covered costs. Download totals in anniversary material indicate adoption, not subscription revenue.¹⁹

The resulting practice had several separable tasks: assemble reusable software; observe what people actually used; supply editorial judgment where automation was inadequate; negotiate with the organizations receiving the traffic. The company could retain the technical work even when the initial product remained incomplete. PasteWall’s long development therefore belongs in the business history, rather than disappearing behind a list of successful launches.

Making AotterGirls: editorial labor and a public identity

AotterGirls began in October 2012, before the news app. Its YouTube channel followed in 2014. This chronology corrects an occasionally compressed story in which the company supposedly created its women-led media operation only after 鄉民晚報 succeeded.²⁰ The company credits Hsieh with proposing the operation and early builders Tseng Li-fang (曾立芳) and Lai Ying-chien (賴盈蒨) with developing it.²¹

By 2018, the department had approximately twelve workers, including five full-time editors and staff in planning, sales, video editing and editorial management. Hsieh told CNA that reliance on student part-timers had become difficult to reconcile with consistent training and quality. From 2015, the operation adopted a more formal talent-management structure: interns practiced presenting, while progression depended on subject knowledge, assignments and independent Chinese- and English-language interviewing.²²

A 2020 campus appearance by presenters Mona and Yu-en gives a more concrete account of the work than the founder’s comparisons with an idol group. They described choosing topics from current developments, adapting them to different audiences and balancing their interests against traffic pressure. Recorded video began with a prepared plan; livestreaming required learning to converse with viewers; podcasting changed the bodily discipline because there was no camera to address. These are different production routines, even when the same personalities appear across them.²³

At that event, a student said she used the channel when she encountered smartphone-software problems. This small, attributable audience testimony is more informative about the service performed than a generic assertion that the brand was “influential.” It shows a viewer using explanation as practical assistance. It is not a customer endorsement of Aotter’s advertising technology.²⁴

CNA reported in 2018 that advertising already provided more than half of revenue and that media had lower margins.²⁵ An August 2026 report estimated media’s contribution at 10–20 percent and Aseal’s at approximately half. These are interview-based estimates, not audited segment disclosures. They help explain why public recognition and economic weight diverge: a presenter is visible to viewers in a way that software integrations, account management and campaign settlement are not.²⁶

The available record does not establish contemporary editor salaries, contractual revenue shares, working hours or a comprehensive staff turnover history. Training systems described by management cannot stand in for workers’ experiences of those systems.

Recruitment, administration and the limits of the workshop model

As the business spread, the back office acquired its own difficulties. In a 2024 customer case published by HR-software vendor MAYOHR, Aotter described managing media, technology and space-and-food operations with different scheduling needs. Its home-built and subsequently purchased tools had become awkward to administer. Administrative director Angela and senior HR specialist Bella discussed attendance, leave, payroll and batch adjustments during typhoons. Bella reported that payroll processing fell from one to one-and-a-half working days to half a day after adopting Apollo.²⁷

This is a supplier’s promotional case, and the reported efficiency gain has not been independently measured. It is nevertheless unusually specific evidence of work otherwise missing from founder interviews. Someone must interpret schedules, resolve exceptions and pay people across unlike businesses. The contemporary slogan of generating high value with few employees does not remove those obligations.

An earlier MAYOHR case, published in 2022, described Aotter using competency assessments alongside interviews. Administrative head Hsieh Wei emphasized communication and problem-solving, and described comparing assessment patterns with interview observations. The account also acknowledged different cultures within different teams. It would be misleading to turn the founders’ preferred traits—curiosity, humor, willingness to learn—into an empirically established description of every workplace in the group.²⁸

AotterTrek and Aseal: building an advertising business

The anniversary chronology places Trek’s formal launch and the establishment of the data-advertising business in 2016. A government entrepreneurship-program profile describes Aseal’s origins in Aotter’s advertising-operations unit. The date of a product becoming a separate operating activity should not automatically be read as the incorporation date of every related legal entity. The current company names Hsieh, Su, Chou Chih-chieh (周志杰) and Tseng Li-fang among Aseal’s founders.^{29, 30, 31}

The early sales problem was partly one of presentation. According to the program profile, customers found statistical charts and distributions too academic. The team responded with simpler tools, including a short link that could distribute visitors among multiple destinations. The same account credits the government’s FAST platform and exhibitions with generating introductions when marketing funds were limited. It supplies a route to customers, but its claims of profitability and hundreds of brand relationships remain promotional interview material.³²

Trek and Aseal perform different, connected functions. A data-management platform organizes audience information; a demand-side platform lets advertisers specify and buy campaigns; a supply-side platform offers publishers’ advertising inventory to buyers. Aotter’s publisher documentation describes integration through web or mobile SDKs, with placements that may be native advertisements, banners or video. Publishers select placements and receive a share of advertising proceeds. The service also offers compensation for contributed data, dependent on its source and quality.³³

This arrangement makes publishers more than distribution outlets. Their implementation work creates the occasions on which advertisements can be shown and behavior observed. The publisher has to place and maintain code; Aotter must support the integration, match compatible advertisements and report results; the advertiser supplies money and creative material. Audience activity is an input to a business transaction without making the audience the direct customer.

On the buying side, Trek’s documentation describes campaign budgets, pacing and measurement of impressions, clicks, video plays and conversions. Its DMP offering describes classifications of interests, geography and audience characteristics. Such labels are model outputs and operational categories, not direct access to an individual’s intentions. A campaign’s apparent precision depends on how signals are collected, how identifiers match and what constitutes a successful conversion.^{34, 35}

The published terms make the economic machinery more tangible. Their September 2020 version distinguishes advertising purchases, publisher placement and paid analytical services. It specifies subscription points for some queries, review of advertising materials, restrictions on fraudulent traffic and conditions affecting payment. The old point schedule should not be presented as a 2026 price list. What it demonstrates is that Aotter sought revenue both from transactions in advertising and from access to analytical tools.³⁶

Capital, subsidies and what the accounts reveal

Gamania’s May 2016 investment announcement disclosed a 15 percent stake and proposed cooperation in content, product development and data applications. This placed Aotter in a network with an established games-and-internet group, rather than simply adding an anonymous financial backer. The contemporary announcement did not disclose a reliable transaction amount.³⁷

There are inconsistencies in the company’s funding story. Its current history calls Gamania the first external funding; its own tenth-anniversary chronology records NT$5 million of angel investment in 2013. Its November 2024 release announced a new strategic investor and said some capital had arrived, but did not name the investor or disclose the amount. The defensible account preserves these dated statements rather than reconciling them by invention.^{38, 39, 40}

A stronger independent record exists for Gamania’s continuing interest. Its reviewed consolidated financial statements for the second quarter of 2026 list a 21.48 percent interest in Aotter, with a nil equity-method carrying amount at June 30, 2026. The same table reports that balance for the comparative periods. An equity-method carrying amount is an accounting balance in the investor’s books; it is not a market valuation, proof of insolvency or a statement that the underlying business has no value. The table does not supply Aotter’s complete ownership structure or a usable stand-alone profit history.⁴¹

Public programs supplied another kind of support. Taipei’s record identifies a 2019 brand-building subsidy project and a 2020 award. These are specific institutional relationships, not evidence that subsidies funded the company’s entire expansion. Taken together with contract work, advertising transactions and minority investment, they show several sources of support operating at different stages. Their relative financial importance cannot be calculated from the available disclosures.⁴²

統一資訊 and the OPENPOINT retail network, 2024–2025

The partnership with President Information Corp. (統一資訊) was announced in November 2024, although later summaries often place it under 2025. The companies connected retail information with advertising infrastructure in a service described as RMN-R. Aotter offered a data clean room installed in the enterprise environment, alongside its retail-media software and advertising network. The release also set a two-year domestic market-share target of 20 percent. A target is not an achieved share.⁴³

President Information had its own strategy. Reporting on its 2025 Lumine launch describes packaging systems and operating knowledge developed within the Uni-President group as services for other businesses. Its retail-media offering supported managed campaigns or self-service buying. The report explicitly distinguished this service from another group company’s Presco RMN. “Working with Uni-President” therefore does not mean Aotter owns or operates every advertising surface in that conglomerate.⁴⁴

The commercial setting changed the kind of audience being sold. An OPENPOINT user might be showing a membership barcode or redeeming coffee, already close to a retail transaction. In an April 2025 interview, Su contrasted such contexts with search advertising. Hsieh discussed building a small organization with very large economic output; the ambition to produce NT$5–10 billion with one hundred people was a future objective, not reported revenue or an achieved workforce.⁴⁵

Aotter’s July 2025 CB LAB tea case shows the proposed working chain. President Information classified retail members; Aotter helped construct the DMP and linked audience categories with Aseal signals through advertising identifiers. Creative material could appear on the OPENPOINT home page, membership-barcode page or coffee-redemption screen. The campaign also included press releases and in-store promotion.⁴⁶

The case reported a 43.42 percent increase in average daily spending per exposed person during the campaign. It did not demonstrate a 43.42 percent increase in CB LAB sales caused by advertising. The published account gives no adequate holdout design, counterfactual or attribution breakdown. Its named client and actual placement are useful evidence that the service was used; the causal performance claim is weaker evidence. The distinction is especially important because simultaneous store promotions could also affect purchasing.⁴⁷

Integration work, identifiers and privacy boundaries

The infrastructure remains dependent on systems Aotter does not control. Publisher integrations include Google advertising tools and the iOS and Android software ecosystems. The company’s public GitHub organization contains mobile SDKs and mediation adapters, making some of the continuing maintenance visible. Its web demonstration includes supported fields for contact information and hashed identifiers. Sample-code capability does not establish that every field is collected from every visitor.^{48, 49}

Privacy therefore cannot be inferred from the phrase “first-party data.” That phrase describes a relationship between a collector and a user; it does not by itself establish informed permission for every subsequent linkage. A local clean room may limit movement of raw records while still enabling consequential matching and targeting. Questions about consent, permitted purposes, retention, access and what leaves the protected environment remain deployment questions.

The technical marketing has also changed. A 2017 government award publication described Aotter’s cross-device fingerprinting technology. Current sales material emphasizes first-party governance and alternative identifiers. The historical description is evidence of the company’s earlier vocabulary and capabilities, not proof that an unchanged fingerprinting system operates in every current product.⁵⁰

Independent standards records offer narrower verification. IAB Tech Lab’s compliance directory lists Aotter’s mobile-app Open Measurement implementation with a September 22, 2026 date. That is meaningful evidence concerning ad-measurement interoperability; it is not a general privacy certification or validation of every claim about GDPR compliance.⁵¹

Platform rules can also constrain a technically functioning system. Apple requires App Tracking Transparency authorization for covered cross-company tracking. Meanwhile, Google’s April 2025 announcement retained Chrome’s existing third-party-cookie choice rather than implementing the anticipated new prompt. Aotter’s “post-cookie” positioning responds to real uncertainty, but a universal, completed disappearance of cookies is an inaccurate explanation of its opportunity.^{52, 53}

Tools that circulated, and tools that stopped

Aotter’s software portfolio included ways to distribute and monetize content as well as to target advertising. The company’s 2023 Sunrise interview explained free short links as an entry into its measurement and advertising services. NewsBuffet, its press-release distribution tool, emerged from the recurring encounter between public-relations senders and journalists looking for material. These are small interfaces placed at existing exchanges of information, rather than attempts to make every user adopt one encompassing platform.⁵⁴

Supr.Link’s product documentation describes campaign parameters and integrations with Facebook Pixel and Google Tag Manager. The short link is consequently a useful example of the company’s double relationship to larger platforms: it offers a locally supplied tool while connecting the customer to external measurement systems. Free access does not mean that the service sits outside the advertising economy.⁵⁵

The expanding portfolio also created unfinished obligations. In October 2023 Aotter cancelled the proposed Supr.Guide shopping-comparison forum, despite saying its core system was complete. It redirected development toward data applications and renamed the planned Trek CDP as Trek Data Suite. The announcement is evidence of a decision to stop and reallocate work, not simply a new name to add to a product catalogue.⁵⁶

The Sunrise account had already described a period of consolidation, with experimentation consuming both labor and budget. Keeping this episode beside the original PasteWall project prevents a retrospective history in which every experiment appears to have been a deliberately chosen step toward the latest market.⁵⁷

AotterClam AI and enterprise implementation, 2025–2026

The enterprise product reused the Clam name for a new problem: operational data scattered across business systems. In the reported case of Toa Lighting (東亞照明), a China Electric brand, project work began in 2025 and a standard version arrived in March 2026. The client initially wanted an ERP replacement. Interviews revealed how heavily existing work depended on the old system; Aotter instead combined exported reports. The August report also named 安源資訊 among paying customers.⁵⁸

Clam’s documentation describes separating reports into tables, relating entities such as orders and products, and normalizing incompatible names, dates and units. Data consultants review each model before launch. Deployment can use client equipment, private cloud or the client’s cloud account. The advertised service contains human analysis and implementation alongside automation.⁵⁹

The site describes user-scoped AI access, query and change logs, and connections to external assistants or private models. These are vendor-described features, not independent security-test results. Keeping a warehouse locally does not by itself prove that an external model receives no data: the actual query path, transmitted context and provider configuration would need inspection.⁶⁰

Commercial delivery also has a partner. In April 2026 Aotter and 福茂數位智能科技, identified in their announcement as a China Electric subsidiary, presented a joint enterprise offering. The partner’s role included consulting, implementation and training, bringing experience from a manufacturer that had itself undergone the work. Clam’s site identifies FullMind as its exclusive general agent. This is a distribution and implementation relationship; it should not be converted into an unsupported claim that China Electric owns Aotter.^{61, 62}

The relationship gives the company something that a general AI demonstration does not: a route into a customer’s routines, an identifiable organization responsible for adoption, and a reference implementation. It also restores a familiar tension. Aotter has repeatedly tried to make reusable products from project work, but each enterprise has different reports, permissions and definitions. How much specialist labor is needed to make a supposedly standard deployment reliable remains economically consequential.

Anonymous testimonials on Clam’s website claim substantial savings and faster reporting, but do not provide enough information to verify those quantities. The named deployment and the published description of the old ERP are firmer grounds for understanding the work than extrapolating an unnamed customer’s savings to the whole market.⁶³

The institution in October 2026

Aotter now presents media, advertising technology and enterprise AI as its principal activities. Its Taipei base remains important: university collaborators, Taiwanese publishers, advertisers, retailers and manufacturers supplied successive testing and commercial environments. A listed Hong Kong liaison office does not establish a large overseas operating network; plans for Japan or Southeast Asia should remain plans unless specific deployments are documented.^{64, 65}

The continuing organization is held together by more than a common logo. Editors and presenters make technology intelligible; engineers maintain integrations and data systems; sales and implementation staff turn technical capabilities into usable services; administrators reconcile differing kinds of employment and schedules. Its history is legible in those recurring tasks and their changing customers. The next stage depends on whether enterprise data preparation can support repeatable paid work without losing the attention to particular workflows that made its first deployments possible.

Explore Aotter’s documented offices on the cross-regional map.

Notes