KOI stands for Knowledge Organization Infrastructure. Metagov’s knowledge-sharing currently occurs through seminars, written outputs and discussion (meetings or chat) and is integral to our development of new projects. Metagov's KOI aims to make the community’s collective knowledge more accessible and deployable across a wider network of people and their inventions.

Knowledge Organisation Infrastructure (KOI) is an open protocol that allows communities to collaboratively manage knowledge on their own terms while remaining interoperable with others. Developed by BlockScience with contributions from Metagov and the Australian Research Council Centre of Excellence for Automated Decision-Making and Society (ADM+S), KOI is designed for contexts where knowledge needs to be contextual, traceable, and machine-readable without forcing everyone into the same database or governance model.

KOI allows different groups to organise, reference, and share knowledge in a modular, consent-based way. It enables interoperability without centralisation, creating a shared architecture for collective intelligence while preserving local control.

Organisations and communities work across various platforms, teams, and contexts. Their knowledge (contained in messages, documents, decisions, processes, and practices) can be fragmented, often buried in chat logs or locked in siloed systems. KOI helps surface and coordinate this knowledge by connecting what communities know, how they govern, and what machines (including AI) can learn or do.

Rather than extract and centralise data, KOI helps organisations to locally standardise how knowledge is referenced and related. It does this through Reference Identifiers (RIDs), which label knowledge objects (such as chat messages, decisions, or datasets) like a library call number. RIDs can point to knowledge without moving it or exposing its contents, enabling systems to refer to, contextualise, and act on knowledge while respecting permissions and provenance.

KOI creates a network of interoperable “nodes” that can signal when knowledge has been created, updated, or removed. This enables dynamic coordination across different systems, agents, and communities. KOI is not about central repositories, but about distributed context.

Metagov’s current instantiation of KOI, “KOI Pond”, combines a knowledge management system (KMS) and RIDs to create a graph database that maps the relationships between knowledge objects, users and groups within Metagov. KOI Pond can be accessed via a large language model (LLM) by using “@KOI-GPT” in Metagov’s Slack channels. 

We believe that KOI supports human and machine collaboration by supporting:

By enabling different nodes to set their own rules and  connect with others via negotiated standards, KOI facilitates the emergence of Artificial Organisational Intelligence: systems where human and machine agents co-produce knowledge, governance, and coordination.

The intention for Metagov’s KOI is that it be guided by Metagov's community and, over time, embody its emergent priorities and self-governance patterns. The aim is for these priorities to surface themselves through intentionally devised practices for incorporating knowledge into our KOI.

The research team hopes to understand:

The architecture of KOI Pond was developed by our partner, BlockScience, who are already using their own KOI. Researchers at ADM+S have also been using their own KOI to construct new ethnographic research tools and practices, expanding on Telescope. Eventually, we hope to observe knowledge-sharing between organizations through peer-to-peer KOI interactions. 

The technical team is currently working on the KOI-net protocol, a standards layer that sets out communication patterns and coordination norms between KOI nodes. KOI-nets are designed to be composable, so organisations can integrate the platforms, tools, and AI agents that make the most sense for their situation. Through this modular design, tasks can be routed to the tools best equipped to handle them, creating a form of governance through configurable delegation. KOI-nets also support permissioning between nodes, making it possible to write and enforce rules on how data is accessed and for what purposes. As development continues, the team is exploring how clear boundaries can be drawn around knowledge objects so that accountability can be embedded into organizational data governance and knowledge sharing practices.

See this FAQ around common questions about the technical infrastructure.

There are many ways that to get involved in Metagov’s KOI development:

1. Experiment with KOI-GPT in Metagov’s Slack

Test the KOI-GPT chatbot within any of Metagov's public Slack channels and share feedback on how you use (or would like to use) KOI. This helps the KOI team better understand community needs.

2. Join project discussions

Most KOI-related conversations take place in the #koi-pond and #govbase-labs channels. Both channels are open to anyone in the community who would like to learn more about KOI or participate in its development.

3. Attend KOI team meetings

The KOI project team meets weekly on Thursdays from 5pm-6pm GMT-4. The meeting link is pinned in the #govbase-labs channel.

We strongly encourage those who are new to the project to attend our open community calls, held during the first meeting of each month. These sessions are an opportunity to learn more about KOI and related research, ask questions, and explore how KOI may fit into the context of your own work.

The remaining weekly meetings are development group working calls, which are open to Metagov members only as observers.

4. Contribute your skills

As the project matures and protocol-level work is finalized, the KOI working group will be seeking contributors from across the Metagov community to help advance KOI. In preparation, the team has begun the process of developing contribution guidelines, creating issue boards, and defining short-, medium-, and long-term goals for the project.

This groundwork will open clear pathways for community members to contribute their own expertise in areas such as technical development, writing and documentation, communications and outreach, and governance and process design.

These developments are still in progress, so watch for future announcements.

BlockScience is continuing to develop the KOI infrastructure. We will be updating Metagov’s KOI in Q3 2025 and will be evaluating its use for onboarding new members. 

KOI is led by Metagov research directors Ellie Rennie and Michael Zargham. 

In addition to the project research directors, several people at Metagov are contributing to the KOI Pond project. Luke Miller is leading the development of Metagov’s KOI. Elianna (Lee) DeSota is the project manager. David Sisson is a domain expert and the project coordinator from the BlockScience team. Brooke Coco is conducting ethnographic research on Metagov’s use of its KOI (pending ethics approval), and Kelsie Nabben is conducting ethnographic research from the BlockScience side. Joshua Tan has played an advisory role and shaped Metagov’s Data Policies. Cent Hosten, Michael Price, and Henrik Axelsen provided valuable input during earlier phases of the project.

Contributions and discussion of this project happen primarily through our weekly calls and online discussions. This project recognizes healthy skepticism as an input to development and governance of ethical infrastructures. The Metagov Board approved two data use policies (here and here), but the work of refining the boundaries of our KOI (IP, ethics, technical) are ongoing. A successful KOI is an infrastructure which can facilitate grounded discourse and a community steered evolution of these boundaries over time.

KOI Pond has received initial funding support through a Metagov DAO grant (for technical development to get the project off the ground). We are currently seeking further funding to enable us to develop tools, interfaces and processes that other groups using KOI may require and/or benefit from.

Do you have questions, want to collaborate, or possibly support this work through funding? If so, here are some avenues to explore:

The “Pond” in KOI Pond is motivated by the dual meaning of Knowledge Organization Infrastructure and the association with a data lake -- or in other words: a data pond. The project also plays with associations to the extrapolation of ideas from Elinor Ostrom’s governance of the commons and land based resource stewardship towards data commons research, as well as fish swarms and interactions in ways that can be traced back to research on collective intelligence.