01 - THE CACHE IS THE CEILING

Every token you generate is stored. That is the bill you keep paying.

A transformer holds every past token as key and value tensors. The cache grows with the conversation, memory tracks length rather than usefulness, and capacity planning becomes a guess. Semantic Gravity Fields replace that cache with a bounded field: fixed footprint, chosen before deployment.

02 - THE CACHE TODAY

Teach the problem before the solution.

Attention needs every prior key and value tensor. Each new token appends another slice, so the cache scales linearly with context length, multiplied by layers, heads, and batch size.

The workarounds all discard: sliding windows drop the oldest turns, summarisation compresses lossily and irreversibly, retrieval re-reads text the model already saw. Each one trades recall for headroom, and none of them gives you a ceiling you can plan around.

FIG. 1 - CACHE GROWTH PER TURN
TURN 1
TOKENS HELD

SCOPE: schematic. Real footprint depends on model dimensions, layer count, batch size, and precision.

03 - FOUR BOUNDED TIERS

One field, four bounded layers, every one of them capped.

Field of meaning

A running representation of the entire conversation, compressed into a fixed number of semantic centers. It never grows past its budget — and the model keeps reading all of it, so a topic raised at the very start stays available a million-plus tokens later instead of scrolling out of reach.

BOUNDED · COMPRESSED
BUDGET: SET AT INIT
Exact-recall pins

The details that have to be exact — names, identifiers, quantities, codes — held word for word and never compressed. The values a wrong answer would turn into a support ticket stay precise no matter how long the conversation runs.

BOUNDED · VERBATIM
EVICTION: NEVER
Recent window

The most recent turns, word for word, over a protected instruction block. The window rolls forward as the conversation moves; the instructions stay fixed.

BOUNDED · LITERAL
WINDOW: CONFIGURABLE
Portable lens

An optional layer you can load into a conversation — a saved persona, profile, or knowledge set: a support team's policies, or what an assistant should already know about you. It shapes how the model responds, sits alongside the live memory without spending its budget, and can be updated at any time.

BOUNDED · PORTABLE
LOAD: OPTIONAL
TOTAL FOOTPRINT = FIELD + PINS + WINDOW + LENS — EVERY BUDGET FIXED
04 - THE FIDELITY DIAL

You set the trade, not us.

One parameter moves the field between lossless and minimal. Everything else about the integration stays the same, so the same build can ship to a datacenter and to a headset.

LOSSLESSMINIMAL FOOTPRINT

Every distinction preserved. Highest footprint.

Merging on redundancy only.

Aggressive merging. Smallest field, gist-level recall.

SCOPE: the dial is qualitative here. Pilot partners receive the parameter, its measured footprint curve, and the recall deltas at each setting.

05 - COMPOSES WITH YOUR STACK

Drop-in, not rip-and-replace.

No retraining

SGF operates on the memory path, not the weights. The model you serve today is the model you keep serving.

Sits beside what you run

Quantization, paged attention, batching, and speculative decoding all still apply. SGF changes what is stored, not how it is computed.

One integration point

A memory interface behind your serving layer. Write the field, read the field, hand the model its context as usual.

INTERFACE: WRITE · READ · PIN · LENS
06 - WHAT WE DO NOT PUBLISH

The mechanism stays private. The results do not.

How centers are formed, scored, and merged is the invention, and it is patent-pending rather than published. What we do publish is everything a buyer needs to judge it: the footprint under a stated configuration, the recall deltas against the same workload on a standard cache, the failure modes we have found, and the conditions where SGF is the wrong tool.

PUBLISHED

Footprint, recall, latency, and the configuration behind each number.

PUBLISHED

Where SGF underperforms: short contexts, single-turn prompts, workloads dominated by retrieval.

UNDER NDA

Parameter-level behavior, integration internals, and the pilot report for your workload.

PATENT PENDING

Center formation, merge scoring, and the pin policy.

07 - PILOT PROGRAM

Run SGF against your own workload.

A scoped four-week pilot: you bring one workload and a memory budget, we return a measured comparison against your current cache: footprint, recall, and the configuration that produced both.

We'll reply to book your discovery call and share the scope sheet and qualifying questions, no deck.

© 2026 SEMGRAVTHE MECHANISM STAYS PRIVATE. THE RESULTS DO NOT.