Memory Readiness Diagnostic.

Tells you which of the five scars your AI memory stack is currently paying for — and how much architectural debt is compounding while you wait.

How this works. Six yes/no questions. Each one corresponds to a memory failure mode we have seen kill AI deployments across 50+ Fortune-500 engagements. Each "yes" = one architectural debt your stack is paying interest on. Score 0/6 means you're a memory architect already; score 5–6 means memory failures are now business risk. Output is yours; nothing leaves your browser.

Have you ever asked your AI assistant a question in a non-English language and gotten a noticeably worse answer than the English equivalent? Multilingual gap
Has your AI assistant ever failed to recall an exact term, ID, or proper noun you know is in prior conversation? Single-store retrieval trap
Are you running an LLM call on every inbound user message — or planning to, as your assistant scales? Per-message extraction cost
Did someone on your team hand-design the folder/category/tag structure the AI uses to organize its memories? Hand-crafted hierarchy
When you ask your AI assistant about something from 6+ months ago, do stale or superseded answers compete with the current ones? No decay layer
If your model vendor changed pricing, exited your jurisdiction, or had its data-transfer adequacy revoked tomorrow, would your team's accumulated memory be at risk? Architecture · Sovereignty gap

0 / 6

0/6 answered

Each question maps to one of the five memory-failure scars (plus one architectural choice) we have paid the tab on across our own builds and the engagements we ship. Score yourself — then decide if it's worth a 30-minute conversation.

Why we trust this score

The six questions map 1:1 to the five scars and one architectural commitment documented in our Memory Layer long-read — each one paid for during the build of our own production memory layer (the substrate behind our agents), and each one validated against a locked eval set with reproducible scoring. Read the long-read for the architecture; take the diagnostic for your stack's status.