Living research case · 1 Nov 2026 # We don’t start from zero. Every week we build on what we already know. [See what we are using →](#this-week)[The three lessons →](/learn/) Public marks Hand-set. Not a live feed. - 26 Oct **No public mark** - 27 Oct **No public mark** - 28 Oct **No public mark** - 29 Oct **No public mark** - 30 Oct **No public mark** - 31 Oct **No public mark** - 1 Nov **Questions opened** An empty day means this page shows no public mark. It is not a claim that no work happened. - Position How we are building. Not a finding. - Open question Asked. Not answered here. - Research note From our research. Not a quotation. - Teaching A lesson to try. Not evidence. - Sample A chart shape. Not a live number. 01 What we use this week ## We don’t start from zero. Every week we build on what we already know. Three lessons Teaching A seat, one shared record, and the rules that bind.[Read the curriculum](/learn/) Roots and add-on Position How we place AI in the work. Not a finding.[Read the position](#roots) Founder pains Research note Notes for teams of about 10 to 100. Not quotations.[Read the notes](#pains) Five measures Sample Definitions for charts. No series is plotted.[See the plates](#measures) ### Open questions, 1 Nov 2026 - Open question ### What should this program reuse? Which research we already hold should the foundation program use, instead of starting over? - Open question ### How do others teach this? Which public examples teach named things, connections and operating rules in a visual way? - Open question ### What does published evidence support? What published evidence supports or weakens a redesign of how work flows, versus AI added on top? - Open question ### How might an AI-rooted organisation operate? What does outside research say about structure, how information moves, and who may decide? - Open question ### What do we already hold? Which notes do we already have, and which gaps still need new research? - Open question ### One curriculum, two readers? How can one curriculum serve people and agents without splitting the meaning? - Open question ### Work that never finishes How does standing work stay visible when it does not close? - Open question ### Which pains do founders name? Which pains do founders of teams this size describe, in their own words? - Open question ### How should the lessons be found? How should the lessons be structured so people and agents find the same meaning? - Open question ### Which sources are safe to cite? Which sources can we cite, and which claims do they support? Draft notes stay off this page until a public teaching version is accepted. 02 Roots and add-on ## Not only adding tools. Outside evidence often points toward redesigning how work flows, not only adding tools. It does not prove our approach. Add-on AI helps too. Our “roots” stance is a position we are building with, not a study result. Position ### Roles and one shared record. When AI sits in how a team decides and remembers, how the work is structured can change. Chat added on top does not do that job. Research note ### Add-on AI helps too. It can help the task in front of you. That help is not a finding that our approach is proved. 03 Founder pains · teams of about 10 to 100 ## Faster tasks don’t help if decisions don’t stick. Research note Five pains from our research notes. Not quotations. Not a measured study of companies this size. - Research note Decisions don’t stick. - Research note Work is forgotten from one tool to the next. - Research note People wait on the person who decides. - Research note How the team works stays hard to point at. - Research note Faster tasks leave the way of working unchanged. Measures we want to watch ## Definitions. No series yet. Sample Each plate is the same empty instrument. The name changes. Nothing is plotted, and a daily start-check feed is not connected. Sample **Research packs filed** How many research packs for this program have been filed. Not a live number. Sample **Questions answered vs open** How many of these questions have a public answer, and how many stay open. Not a live number. Sample **Insight drafts vs accepted** How many notes are still drafts, and how many have been accepted. Not a live number. Sample **Research per weekday vs target** Weekday research completed, set beside the target for that work. Not a live number. Sample **Work Window snapshots** Whether a public snapshot of the work window was taken. Not a live number. Later ## The whole experiment. An optional subplot for the full experiment can live here later. It is not on this page yet. Not started. No dates and no conclusion. Meanwhile ## Read the lessons while this stays open. Teaching The curriculum is the practice. This page is the case behind it. [Open the three lessons →](/learn/)