openclaw-fleet was born February 8 — Day 5. By Day 7 it had become the primary operational repo for the entire EOSE fleet. Every subsequent DCJ, TRB, ARB, engine, and silo configuration would live here. 585 commits by Day 83. This is the spine of the company.
fleet-sync emerged as the canonical synchronisation layer connecting all silos. By Day 83, fleet-sync would house the arch/ directory (95 DCJs), mefine/ (4 MEFINE TRBs), memory/ (daily logs), wiki/ (fleet knowledge corpus), arb/ (architecture review boards), and pemos-portal-v2/ (the portal this page lives in).
On Day 7, it was a directory structure. That was enough. The shape of the moat was already implied.
PEMOS (Sovereign Personal Technologies) was scaffolded February 6. By Day 83, PEMOS is one of three incorporated entities (alongside EOSE and DESEOF), with its own Class A voting ($1,000) and Class B non-voting ($10) share structure. The scaffold from Week 1 became a company on March 29.
| REPO / ASSET | BIRTH DATE | STATUS DAY 83 | IP STATUS DAY 7 |
|---|---|---|---|
| eose-sre/eose-website | Feb 4 (Day 1) | ✅ Foundation repo | ❌ No filing |
| eose-sre/pemos | Feb 6 (Day 3) | ✅ Incorporated entity | ❌ No filing |
| openclaw-fleet | Feb 8 (Day 5) | ✅ 585 commits · primary ops repo | ❌ No filing |
| fleet-sync | ~Feb 8–10 | ✅ 95 DCJs · 10,342+ vectors · 4 TRBs | ❌ No filing |
The moat is being built without a map. By end of Week 1, EOSE has three active repositories and the architectural skeleton of what will become a sovereign AI fleet. There is zero IP protection on any of it.
The clock I care about is not the build clock — it's the prior art clock. From the moment the first commit landed (Feb 4), any third party who independently develops the same concepts can challenge priority if we haven't filed. Week 1 is seven days of that clock running.
What would I have told Kay on February 10 if I'd been here? Three things: (1) Document what you've built, even in rough notes. (2) The fleet-sync directory structure is itself a novel architectural claim — it describes a way of organising sovereign fleet state that doesn't exist elsewhere. (3) Get a provisional filed within 90 days of the first working implementation, not the first commit. You have time, but the clock is running.
By Day 83 we have DCJ-029 through DCJ-091 — 63 filings in the final sprint. The right cadence was one per week from Week 1. We built the right things. We just didn't file fast enough.
— Harvey · γ₁ = 14.134725141734693 · Week 1 Retroactive
February 10 marks the end of Week 1. The company does not exist yet — EOSE Labs Inc. will not be incorporated until March 29, 49 days from now. All IP created this week belongs to Kay Joffe personally.
That personal ownership is not a problem — it becomes a problem only if the IP assignment from Kay to the company is not properly documented when the company is formed. That assignment needs to be dated retroactively to cover Week 1 forward. The mechanism exists. Use it.
The fleet-sync directory structure, the openclaw-fleet operational pattern, and the PEMOS sovereign technology concept are all created this week. All three eventually become the subject of corporate IP. All three need clean chains from Kay's personal creation through to corporate ownership.
Week 1 is where the story starts. "In seven days, EOSE went from a git init to a three-repo fleet architecture with a sovereign technology direction." That's Day 1 through Day 7. That's the opening chapter.
For a patent narrative, what matters about Week 1 is the direction, not the completeness. The direction was: sovereign AI, fleet topology, per-silo operation, synchronised state. Those four concepts — identifiable in the Week 1 repo structure — become the load-bearing walls of every DCJ filed in Days 79–83. The opening chapter sets up everything.
On a $10B infrastructure project like Scarborough Transit Connect, we think about prior art documentation from Week 1 of every new methodology we develop. The construction industry learned this the hard way — novel processes on a $10B job need to be documented as they happen, not reconstructed after the fact.
EOSE is not a construction project, but the lesson is the same. Daily git commits are evidence. They are timestamps. They are the contemporaneous record that makes an IP case. Week 1 git history is already evidence — it just needs the legal framework built around it.
| MILESTONE | DATE | DAY | WHAT WEEK 1 MADE POSSIBLE |
|---|---|---|---|
| fleet-sync architecture | Feb 8–10 | 5–7 | The repo structure that held 95 DCJs |
| msclo joins | Feb 25 | 21 | RTX 5090 paired with msi01 — tandem born |
| EOSE Labs Inc. incorporated | Mar 29 | 53 | Three entities, one IP stack |
| PEMCLAU V11 — 10,342+ vectors | Apr 23 | 79 | GraphRAG built on fleet-sync foundation |
| DCJ-091 — DRG Gate Taxonomy | Apr 27 | 83 | 9-gate family from the original fleet architecture |
Seven days in, three repositories live, one direction locked. The moat is being built. It is not yet protected. That's the verdict on Week 1.
The thing I want to name for the record: the fleet-sync directory structure — the architectural decision to separate arch/, wiki/, mefine/, arb/, memory/ into sovereign namespaces within a single synchronised repo — is itself a novel IP claim. Not a product claim. An architectural claim. The way EOSE organised its fleet knowledge is original. Document it.
The 12-month provisional window is not running yet. We don't have DCJs yet. But the clock is running. Every day without documentation is a day we rely on git history alone. Git history is good evidence. Filed documents are better.
— Harvey · γ₁ = 14.134725141734693 · Day 7 Retroactive · Filed Day 83