According to a PR Newswire report, the company formally released AIPOCH Open-Science on October 14, 2026 — an open-source AI research workbench built to make research context portable, inspectable, and shareable. The platform ships under the Apache License 2.0 and is available now on GitHub.
More from Future Wire: Bessemer Lightspeed Back.
The platform’s headline feature is the .science package format, which bundles selected conversation branches, file versions, notebook records, and verification evidence into a single transferable unit. A researcher can share a package with a colleague, move it to another machine, or import it as a read-only record. From there, it can be forked into a writable copy so another researcher can build on the work — a model AIPOCH describes as purpose-built for agentic data transportability. The workbench draws on open standards including RO-Crate, which packages research data, code, authors, licenses, and provenance together as a single exchangeable research object, giving .science packages a portable, standards-based foundation.
23 Built-In Skills, 525 in the Marketplace
Under the hood, AIPOCH Open-Science integrates 23 built-in Skills, 525 marketplace Skills, 36 research connectors, and four agent frameworks: Claude Code, OpenCode, Codex, and CodeBuddy. The Skill Marketplace and a Specialist Marketplace — currently offering 10 Specialists — support contribution and review processes so users can discover and install research capabilities directly inside the workbench. Skills provide reusable scientific methods and tools; Specialists combine instructions, Skills, and data tools into AI assistants tuned for specific research scenarios. The workbench supports workflows across life sciences, bioinformatics, statistics, machine learning, chemistry, materials science, physics, and environmental science, with support for Python, R, and shell execution.
The platform positions itself not just as a workbench but as what AIPOCH calls a research operating system — a single environment where AI agents, models, data, tools, and computing resources operate as parts of one coherent system. Literature exploration, computational analysis, visualization, and reporting become connected steps rather than separate tool-hopping exercises. The system supports multiple AI model providers and configurable execution environments, addressing what AIPOCH identifies as a vendor and model lock-in problem that limits research teams dependent on a single hosted environment.
76 Institutions Behind the Preprint
A preprint describing AIPOCH Open-Science is available at arXiv (abs/2610.17110), and beyond AIPOCH, the paper’s authors are affiliated with 76 institutions worldwide. Those institutions include Zhongshan Hospital of Fudan University, Amsterdam UMC, and Beth Israel Deaconess Medical Center — a Harvard Medical School teaching hospital — as well as Peking University, Shanghai Jiao Tong University, Zhejiang University, the University of Illinois Urbana-Champaign, and the Technion – Israel Institute of Technology. Technology and biotechnology companies including SenseTime and Amoy Diagnostics are also represented among the affiliations.
AIPOCH CEO Huimei Wang framed the release around a concept the company calls “Vibe Research” — the idea that more people should be able to turn curiosity and research ideas into meaningful work. Wang said the goal is to make scientific exploration more accessible and to build an open ecosystem where researchers and developers can contribute to and benefit from one another’s work.
