How it works

From pip install to grounded answers in four steps.

Read each document into the model's memory once — one forward pass, no training — then serve every question about it from that memory, for every workload on your fleet.

The problem

Document workloads re-read the same text on every query.

Every question pushes thousands of context tokens back through the model — fleet GPU time spent re-reading documents that were already read an hour ago, and yesterday, and the day before.

01

Pay-per-read forever

Every query re-prefills thousands of document tokens. The re-reading burns fleet capacity in proportion to traffic.

02

Latency grows with context

Time-to-first-token climbs with every token stuffed into the prompt — and it's your SLO that absorbs it.

03

Prefix caches don't survive production

The engine's built-in cache is keyed to exact token prefixes and evicted whenever the pool churns — so clients re-ship whole documents to rebuild state that existed an hour ago.

The approach

Read the knowledge once. Serve it from memory.

Engram Smart CAG reads each document through the model once and saves the model's internal working memory of it; durable, resident on your GPUs, and addressable by document id. A flexible, composable document base cache, made permanent. Same model, same answers; the reading work is done once instead of thousands of times.

$

Read-once economics

Onboarding is sub-second to single-digit seconds per document, once. After that each answer prefills dozens of tokens instead of thousands — measured 91× less re-processed text per question.

Whole-document grounding

The saved memory covers the full document, not a chunk. Matches the quality of re-reading the entire document, where chunk-RAG loses context.

Two serve modes

Attend the whole document's memory (CAG) — or, when storage is tight, a training-free router loads only the slice that answers the question (QRC). Same measured accuracy.

The pipeline

Four steps, one command.

1

Install

pip install the plug-in — a stock vLLM connector. No fork, no new engine; wire it to your tenant API or document store.

2

Read once

One frozen forward pass per document saves its KV memory. No training, no fine-tuning, ever.

3

Retrieve & route

Bring your own retrieval or use the built-in hybrid stack — either way it just picks document ids; a training-free router can narrow to the answering slice.

4

Answer

The model answers from the resident memory — only the question itself is prefilled, dozens of tokens. Grounded, sourced, first token in ~30 ms.

🔒

A data story you can sell.

Engram runs inside your VPC with no public endpoint. Your customers' documents never train a shared model, never leave your perimeter, and deleting a memory removes the document from serving — data-residency guarantees you can put in a contract.

Get in touch