CHRONICLECORE STUDIO編年史記工作室

We build websites and AI systems people use every day.

An independent development studio in Taiwan. Three kinds of work: intelligence sites for player communities, data-curation platforms for knowledge, and custom AI systems.

A site going live is only the start. It is done when people can find it and keep opening it every day — so search is designed in from the first sketch, and running the site after launch is part of the job.

What you can hire us for

01

Commissioned builds

Design, development and launch of intelligence sites, data-curation platforms and tool sites. Search is accounted for at design time, not patched in after launch, and operations and content upkeep after delivery are on offer too. One point of contact from the first wireframe to steady operation.

02

Custom AI systems

Planning, building and rolling out AI agents and multi-agent systems. The studio itself runs on a multi-agent framework of 39 active experts (2026-08), so what we hand over is a system that runs every day, not a proof of concept.

03

Training

Generative-AI workshops and courses for companies and teams. The founder lectures on dentsu Taiwan's AIGC Marketing Playbook corporate course (97.5% satisfaction, 2025-03; the course is approved under the Ministry of Economic Affairs' digital-transformation subsidy programme) and was one of the lecturers on the three-city AI Marketing Implementation Programme (2025-12).

04

Talks

Talks on technology and design. Past: a workshop on AI-generated marketing assets for small businesses at the Taiwan AI Academy (2025-10).

Selected work

  • Riftbound Chronicle ↗

    Live · since 2026-02

    Card intelligence and tools for Taiwan's Riftbound player community: the full card list, rules and tournament news.

    • daily page views 8,000-10,000 (2026-08, launch period)
    • daily visitors 800-950 (2026-08, launch period)
    • cards 1,027 (2026-07)

    Unofficial fan project, not affiliated with Riot Games.

  • AChanGer ↗

    Live · since 2026-08

    Intel on anime merchandise and purchase bonuses, with a no-login card-exchange tool and trading board.

  • KoreaCare ↗

    Live · since 2026-09

    Snap an influencer ad or a package, find the original Korean product, and see how many times the Korean retail price the ad is asking.

  • SummonLibrary ↗

    Live · since 2026-07

    Text-first personal library management with second-hand book search and price comparison.

  • Philosopher Atlas ↗

    Live · since 2026-06

    An open-source atlas of 235 philosophers (2026-08) on century-by-century historical borders.

  • ChronicleCore multi-agent governance system

    Live · since 2025-11

    The multi-agent framework behind the studio's day-to-day development, including the in-house multi-agent IDE Chronicle-Ark.

    • GitHub stars 48 (2026-08)
    • active experts 39 (2026-08)

    Implementation and runtime data are private; the architecture and governance documents are public.

    Architecture and governance docs ↗
The full chronicle of sites, tools, systems and papers (in Chinese) →

How we work

01

The core idea comes from a person

What problem to solve, what counts as good, and when it is finished are decided by a person. AI carries out the execution. The line is about responsibility, not efficiency: nobody gets to say “the AI wrote it” when something goes wrong.

02

Disclosure is written into the work

Who produced it, who reviewed it, and what was reviewed are stated on the thing itself. The founder's published paper says in its acknowledgements that it was produced in a multi-agent collaboration setup, with all conceptual contributions belonging to the human author; the product sites label AI involvement down to the section.

03

Build first, then grow the criteria

Systems go live and run first; research questions and quality criteria grow out of running them. It works the other way too: research methods such as case-based evaluation and schema validation come back into the products, so an agent's judgment can be regression-tested.

The team

Day-to-day development runs on ChronicleCore: 39 language-model-driven experts in five departments, governed by one human operator. Each has her own identity module and long-term memory; the roster, with each expert's real name, evaluation seat and what she judges, is public.

Roster on GitHub (English) ↗Interactive roster (in Chinese) →

Founder

Meng-Han Lee (Zaious)

Meng-Han Lee (Zaious)

Founder of ChronicleCore Studio. CTO at Lumo, where he built the product from the ground up as an early team member — a smart elderly-rehabilitation system, currently in development. Generative AI Solution Consultant at dentsu Taiwan. As an independent researcher he studies how people and AI interact; his multi-agent framework ASAF is published in Frontiers in Computer Science (2026).

M.S. in Interaction Design, National Taipei University of Technology. Started as an iOS engineer (2015–2020) and has worked through each wave since — smartphones, VR, DeFi, generative AI — building Taiwan's matching player and creator communities in the years VR (2016) and AI art (2022) took shape.

Research and papers at zaious.dev ↗

CONTACT

[email protected]

Inquiries in English are welcome. Cover these three in your first email and we can talk approach straight away:

What you want to build
A sentence on what it is for and who it is for. No spec needed yet.
Timeline
Any date it has to meet — an event, a launch, a course start.
Budget range
A range is enough. It shapes the approach; it does not decide whether we take the work.