Vision & Roadmap

The Semantic OS

An operating system nobody planned

An operating system emerged with no master plan

Sixteen people across 500+ repos built 442+ capabilities over the past year — and no one drew a blueprint. Seen from above, the pieces line up like a coherent operating system.

The whole deck defends one claim: this shape is real.

16
people
442+
capabilities
500+
repos
0
master plans

The claim is anchored in a real, dated document

It comes from ramparte/amplifier-os-vision — a single-author strategic-analysis repo (Sam Schillace, 6 commits, April 2026). Its central artifact, the-semantic-os.md v0.2, runs 1138 lines, built over ~5 days.

A versioned primary source — so we can trace every number back.

1
author (Sam Schillace)
6
commits
1138
lines (v0.2)
~5
days, April 2026

Nobody coordinated, so the same need got solved again and again

Without a master plan, the team reinvented the same capabilities in parallel — four independent M365 connectors, six separate memory systems, three schedulers that never talk to each other.

Wasteful — until you read the redundancy as a signal.

4
independent M365 connectors
6
memory systems (3 personal vaults)
3
non-coordinating schedulers

The vision names the shape it found: three layers

The document structures the emergent OS as three layers, described as different in kind — a Substrate underneath, a Self-Model in the middle, and an Elevation layer on top.

Layer 2 is where the argument is headed — hold onto it.

Layer 1
Substrate
The plumbing.
Layer 2 · highest-leverage gap
Self-Model
The system reasons about itself.
Layer 3
Elevation
Raise the human's abstraction level.

Read the redundancy as a demand signal

When eight projects that never referenced each other all reinvent the same propose → judge → refine loop, that isn't waste — it's the team voting with its code for a native primitive the platform should provide.

That single move turns the cost into a direction.

Propose
A candidate is generated
Judge
It's presented and scored
Refine
Repeat until convergence
Reinvented in 8+ independent projects

The roadmap isn't invented — it's read off the code

The document tabulates nine missing system-level capabilities, derived from the redundancies. Leading them is the Self-Model — Layer 2, called "the highest-leverage gap in the current ecosystem," still mostly unbuilt.

The gaps are empirical, not wished for.

9
missing system-level capabilities
L2
Self-Model — highest-leverage gap

The redundancy is the roadmap — and an independent vision agrees

A year earlier, bkrabach's Cortex Platform vision — three documents totaling ~100KB from March 2025 — arrived at the same answer. Section 10 calls the convergence "remarkable."

Independent convergence is the strongest signal the roadmap is right.

3
Cortex documents (~100KB)
2025
March — a year earlier
2
independent visions, one answer
The North Star

"Maximize the value of outcomes per unit of human attention, while holding human attention constant or decreasing."

The objective function — Section 8, the-semantic-os.md. Build the substrate the team already asked for, and steer by this.

Sources

Sources & Research Methodology

Data as of: April 2026 · Primary source: ramparte/amplifier-os-visionthe-semantic-os.md v0.2 (1138 lines)

Commands run (re-verified in a fresh session):

Gaps & caveats: The 442+/500+ aggregates are the doc's own figures from the team_knowledge base, not independently recounted in this research. The member count grew 15 → 16 between docs; reconciled in attribution-audit.md:287 ("With payneio and dluc included, the true number is 16"). Cortex details are the vision doc's own internal claim; a separate Cortex source repo was not located in this research.

Primary contributor: Sam Schillace — sole author of all 6 commits.

Feature status: Working Document (v0.2) — Active

More Amplifier Stories