The Cambridge Autonomy Reading Group is the directory's academic anchor. It is not a startup, not a company, and not in the strict sense an institution. It is a termly working group of researchers from three UK institutions who meet to argue about agentic systems and circulate a reading list afterwards. The Bulletin has tracked the group since the second cohort and has cited their reading lists more often than any other academic-side artifact in our coverage.

The group is convened by Dr. Imogen Fairley, a computer scientist whose published work is on multi-agent verification, and a rotating group of co-conveners drawn from her department and two adjacent ones. The membership is small and informal. The group does not publish a journal. The reading lists it produces are short — usually fifteen to twenty papers per term, with two to four longer essays alongside the technical work. The lists circulate by email and are not, formally, public artifacts. The Bulletin has been given permission to reference specific items in our coverage.

What makes the group editorially useful is the curation. The reading lists are unusually opinionated for an academic working group. They mix recent technical work on multi-agent systems with older papers from cybernetics, organizational theory, and the early literature on workflow management. They are not exhaustive. They are not balanced in the sense academic surveys are usually balanced. They read as positions: as the working group's collective view of which intellectual ancestors actually inform the current shape of the field.

The Bulletin has cited reading-list items in our essays on UI patterns for agentic products, on agent identity, on the operating-system metaphor, and on the question of which abstractions survive a platform shift. The group's reading lists have shaped the Bulletin's own essay programming in ways the editorial team has been explicit about. Our cornerstone essay on what Web4 means draws on three pieces the group circulated in 2024. Margot Halloran's comparative work on Web3 and Web4 draws on a fourth.

The directory entry is here because the Bulletin's commitment to track the structural elements of the Web4 thesis includes its academic substrate. Most agentic-AI publications cover universities only when a faculty member founds a company. The Bulletin has chosen to keep room for academic work that does not have a commercial output, because the long-term durability of the category depends partly on whether it can produce real intellectual scaffolding rather than only product narratives.

The group has, to its credit, refused to commercialize. There is no nonprofit shell, no sponsorship arrangement, no public list of corporate observers. Several of the directory's other entries — entries we do not name here — have approached the group about collaborations and been declined. The Bulletin's view is that the group's value comes partly from that refusal, and we have been careful not to push for more access than the convenors are willing to give.

We expect to continue tracking the group continuously, with the same access posture we have had so far. Updates to this entry will be made when the group's structure or output changes substantively, or when one of their public reading-list items becomes structurally important to the broader Web4 conversation.