Context Engineering

What should your agent remember - and who decides?

Memory types, governance, retention, and the canonical rule for when memories conflict.

Memory is not storage. Memory is identity. What your agent remembers shapes who it becomes - and what it forgets determines what it can never be. Most agent systems treat memory as a technical problem. It's a governance problem.

This guide gives you the framework for designing memory that serves people - including the right to be forgotten.

What You Get

Memory Type Taxonomy

Operational memory, identity memory, relationship memory, and decision memory. Each type has different governance needs. Treating them the same is the first design mistake.

Memory Governance Framework

Who can write, read, modify, and delete each type of memory. Authority without governance is just access control. Governance without authority is just policy.

The Canonical Rule

One sentence that defines which memory source wins when memories conflict. The most important sentence in your entire memory architecture.

Retention Policy

What stays, what expires, and what must be deletable on request. Includes the right to be forgotten - not as compliance, but as sovereignty.

What's Inside

1

Memory Types

"What kinds of memory does your agent hold - and are you treating them all the same?"

2

Memory Governance

"Who has authority over each memory type - and what constrains that authority?"

3

Session Protocols

"How does memory persist across sessions without accumulating drift?"

4

The Canonical Rule

"When two memory sources disagree, which one wins - and why?"

5

Retention & Forgetting

"What is the right to be forgotten - and does your architecture honor it?"

Who This Is For

AI Developers

Building agents with persistent memory. The architecture decisions you make now determine whether memory serves the user or surveils them.

Privacy Engineers

Agent memory creates new categories of personal data. This framework gives you governance that maps to privacy principles, not just compliance checkboxes.

Product Teams

Your agents remember things about your users. The question is whether you've designed that memory with the user's sovereignty in mind - or only your feature roadmap.

$49

PDF download. Design memory for as many agent systems as you need.

Buy the Guide

Includes the memory type taxonomy, governance framework, session protocols, canonical rule template, retention policy, and 10-minute memory diagnostic.

Go Deeper

The guide gives you the memory architecture. If you want the full trust framework around it, two options:

Trust Architecture Blueprint

Memory is one of four pillars. The Blueprint shows how identity, memory, governance, and refusal rights work together. $49.

See the Blueprint

Discovery Call

Free 30-minute call to discuss your memory architecture and figure out next steps.

Schedule
Chat with Echo

Hi, I'm Echo — an AI assistant for evoked.dev. I can answer questions about Erin's work, services, and projects. What would you like to know?