A Field Guide to the Distributed AI Economy
About this guide
This is a field guide, in the naturalist's sense: a book you take into unfamiliar terrain to learn what things are, how they relate, and what to expect from them. The terrain here is a shift underway in enterprise AI, away from renting intelligence through vendor APIs and toward owning it as governed infrastructure, and toward an economy in which units of AI work, the workers that perform it, the organizational context that orients them, and the checks that verify them can all be published, installed, and exchanged.
Our primary source is a whitepaper: The Distributed AI Economy: Intelligence Hubs, Frames, Cogs, Ops, and the Accountability Plane (OpenTeams, August 2026, Revision 9), written by Travis Oliphant, the creator of NumPy, a co-founder of Anaconda, and CEO of OpenTeams. Revision 9 opens with a signed personal statement, "Why I Am Writing This," and closes with a credo in the same voice. It is no longer quite "a company says"; it is "a named person argues, and one company he leads is acting on the argument." It also discloses its own interest before any reader can raise it: "This paper is not a product pitch. OpenTeams, which I lead, participates in this ecosystem as one contributor among many, and I describe its choices only where they illustrate the pattern."
That changes the posture this guide adopts. A vendor whitepaper is a document to discount; a signed argument is a document to test. So this guide tests it, holding two things in view. A disclosed interest is still an interest. And the paper's closing promise that "other materials I produce will show how some of that value can accrue to specific, investable organizations, OpenTeams among them" is a promissory note worth noticing: the investable case is being made elsewhere, on terms this document does not have to defend. What remains here is a serious piece of systems thinking whose claims deserve attribution, whose evidence deserves inspection, and whose bets deserve to be named as bets.
Revision 9 also grades its own maturity, in a three-row table separating what exists today from what is in active development and what is still a thesis. This guide has kept those three tenses apart from its first edition (shipped software, hardening standards, forecast economy) and can now check its ledger against the paper's own (Section 12, FAQ 15).
Three conventions keep those registers separate throughout:
Two smaller conventions also recur: Precedent boxes carry historical parallels (NumPy, shipping containers, TCP/IP), and In practice boxes carry illustrative worked examples, which are invented for teaching and marked as such.
The map
The guide is organized as five parts. Each section stands alone; read in order for the full argument, or jump to what you need.
A reading map of the guide. Part I, The Problem, covers rented intelligence and context as capital. Part II, The Execution Layer, covers Frames, Cogs, and Ops, plus the cross-cutting accountability plane of Guards, Gates, and Tracks. Part III, The Ground It Stands On, covers the Intelligence Hub, the open-source commons it is assembled from, and the products that make the Hub usable. Part IV, The Economy, covers the four-class market, why openness compounds, and the decade ahead. Part V is reference material: the stack drawn on one page, a glossary, an FAQ, and further reading. Each section builds on the ones before it; readers can also jump directly to the part that answers their question.
View as text diagram
PART I — THE PROBLEM 1. Rented Intelligence why consuming AI as a tenant fails 2. Context Is Capital the asset that leaks away PART II — THE EXECUTION LAYER 3. Frames organizational context as a versioned artifact 4. Cogs AI workers you can audit 5. Ops jobs you can install 6. Guards, Gates, Tracks the cross-cutting accountability plane PART III — THE GROUND IT STANDS ON 7. The Intelligence Hub the perimeter where it all runs 8. Assembled From the Commons open source, Nebari, and Nebi 9. The Doorway the products that make the Hub usable PART IV — THE ECONOMY 10. The Four-Class Market what trades, and on what terms 11. Why Open Compounds network effects and open standards 12. The Decade Ahead timing, value capture, open questions PART V — REFERENCE 13. Drawing the Stack the whole architecture on one page 14. Glossary 15. FAQ 16. Further Reading
Readers who came for the economics can go straight to Part IV; each of those sections opens with the prerequisites it assumes, linked back to where they are taught.