Things you make, and everything that becomes of them.
Later layer · deferred by design
GEMS is a later layer of the factory. Artifacts are already sentences in the GIMS ledger;
GEMS is the read-view that keeps their identity intact across platforms, formats, and time.
It waits for the factory to produce artifacts worth mapping: runtime first, then the ledger,
then the foreman. See the blueprint
for the build order.
GEMS is the organ that will hold the semantic truth about artifacts: the
things you produce, every place they live, every derivative they spawn, every variant
that gets re-cut for a new audience or a new platform. The aggregate source of truth that
the patchwork of platforms cannot, and will not, give you.
Where GIMS records the verbs of operational reality, GEMS will read them through the lens
of what was made. Artifacts are first-class entities; their identity survives
across platforms, formats, edits, and time.
You make a thing. What was it called when you started? Where did the working file
go? What clips were cut from it? Where did those clips end up posted? What captions did the
vertical version get? When that platform changed its rules, did you re-upload the variant
that complies? Five seconds in and you're already losing.
Every platform owns its own metadata silo and tells you nothing about the others. Every
tool in your editing stack invents its own filename. Every collaborator has their own naming
scheme. Semantic drift is the default state. A year later, you can't find
the source for half the things you've shipped.
A Worked Example
One video, lived through reality
Below is the lineage of a single piece of long-form video content, traced through what
actually happens to artifacts in the wild. GEMS will keep this trail; the absence of GEMS
is why most creators can't.
GENESIS
A 48-minute video on "how production scheduling actually fails in industrial labs."
Working title scheduling-fail-v3-FINAL2.mp4.
Origin recorded with the prompt, the brief, and the date.
DERIVATIVES
Cut into four shorter pieces (10 min, 4 min, 90s, 30s), two vertical
reformats, one transcribed blog post, one podcast audio strip. Each derivative is a
noun in GIMS terms, but GEMS will see them as a constellation around
the original.
DISTRIBUTION
The 90s vertical posted to TikTok. The 4 min cut posted to YouTube. The blog post to
the company site. The podcast strip to Spotify. Each has a different upload date, a
different caption, a different platform-side identifier. GEMS will track all
identifiers as references to the same underlying artifact identity.
DRIFT
Six months later: a new platform's algorithm change. A regulatory caption update. A
collaborator wants to repurpose the 4 min cut for a sales deck. GEMS will know where it
lives, what version it is, and what variants exist. Nothing is lost. Nothing is
re-rendered. Nothing is re-searched-for.
REUSE
A year later, GEDS will be able to surface a hypothesis: "three of your top-performing posts share
the same underlying clip. There is a fourth derivative you haven't yet cut that fits
this pattern." The reuse opportunity exists because the constellation exists.
The Mechanism
Identity that survives the platform
The hard problem is artifact identity. The design solves it by making identity a function of
origin, derivation chain, and content fingerprint, not any specific filename,
platform ID, or storage location.
01 · IDENTITY ≠ LOCATION
An artifact's identity is not its filename, its platform ID, or its URL. Those are
references. The identity is the tuple of origin, derivation chain, and
content fingerprint. Move the file, rename it, re-encode it: the identity is
preserved.
02 · DERIVATIVE CHAINS
Every derivative declares its parent and the transformation that produced it. "This
is a 90-second vertical reformat of the 4-minute cut of the 48-minute original."
Six steps down, you can still trace the line.
03 · MULTI-PLATFORM IDENTITY
One artifact, many platform-side IDs. GEMS will map all of them to a single internal
identity. When you ask "how is the 90s cut doing across all platforms?", you
get one answer drawn from many sources.
04 · REUSE FUEL FOR GEDS
The constellation is rich training data for GEDS. "This series of edits worked. This
combination of platforms produced the best return. There is a derivative pattern you
haven't yet exploited." GEMS will be GEDS' favorite reading material on the creator side.
Where Artifacts Live
Every platform, one identity
Real artifacts don't live in one place. They scatter across distribution surfaces, each
with its own identifier, analytics shape, and posting rhythm. GEMS will unify them on the
identity side without flattening the differences on the platform side.
▶
YouTube
long-form · evergreen
♫
TikTok
vertical · feed
◯
Instagram
reels · carousel
⤳
Reddit
subreddit · thread
§
Blog / Site
prose · embed
The Defining Capability
500 videos, all still findable
A creator's working backstory: five hundred recoverable videos. Drives full of
unlabeled edits. Captions on six platforms, none of which agree with each other. The
"where did this clip come from" question loses an afternoon every time it's asked. The
"do we already have a piece on this topic" question gets answered wrong, because the
archive is opaque to its owner.
With GEMS, the constellation becomes the asset. Every artifact you've ever made
is a queryable entity with full lineage to its source, full visibility into its
derivatives, and full distribution context across the platforms it touched.
Reuse stops being an archaeology problem. Catalog stops being a graveyard. The body of
work becomes a body of work, not a junk drawer.
Where It Fits
GEMS in the graph
This is why GEMS can wait: it is a specialized read view over GIMS. Artifacts are already
sentences in the ledger, so GEMS adds no second source of truth — it is the artifact-shaped
slice of the substrate, with first-class write paths for derivation events and
cross-platform identifier mapping. It's the organ that will turn a noisy production
pipeline into an aggregate, addressable body of output.
GEMS — Artifact Semantics
Identity that survives the platform. Lineage that survives the edit. Reuse that survives the year.
↔ GIMS
The institutional ledger, already in the field. Artifact lineage lives there as sentences; GEMS is the artifact-shaped read view.
→ GEDS
The five-mode relevance engine. The constellation is fuel for it: reuse opportunities, drift patterns, derivative gaps.
← GOMS
The intent compiler: goal to campaign to sprint to task. Tasks that create artifacts will register them at the point of creation.
↔ GRAMS
The market membrane, itself a later layer. Distribution channels feed back platform IDs; outgoing proposals can cite specific artifacts.
→ GONS
The foreman and the command room. Stale-derivative alerts and reuse triggers will route through GONS-Core and surface in the Console.
GEMS is the layer that will make a body of work compound instead of disperse.
The library is only an asset if you can read it.