Use cases

Six shapes of agent memory

One vault, scoped per agent. Each verticale ships with a full article, a diagram, and a Python snippet you can paste into your own codebase.

Coding agent

The conventions your team agreed on, remembered. The refactor decisions, surfaced. One folder per editor, plain Markdown on disk.

Read the full article

Deal stages, objections, contact preferences. One vault per SDR, audit trail per contact, valid_at on every fact.

Read the full article

Past tickets, resolution channels, preferred contact path. A returning customer is recognised before the first message.

Read the full article

Recurring mistakes, learning style, the explanation that landed. One profile per student, adapts at every lesson.

Read the full article

Patient context with audit logs, lawful-basis tags, on-prem or EU residency. Designed for GDPR and NIS2.

Read the full article

Long-term timeline of the user. Color preferences that shift, jobs that change, goals that evolve. Memory that compounds.

Read the full article

Want a verticale we haven't covered?

Email me with the shape of memory you need and the agent you're using. If it's a recurring pattern, it ends up on this page.

Or join the early-access cohort to get the Python SDK first.