Resources
The SAIHM manual
Memory That Works for the Whole Organization: a decision-maker’s guide to SAIHM and SHM. What it is, how it protects your data, how agents share memory safely, and how to defend the choice in a security review.
39 pages, 1.7 MB, tagged for screen readers.
How to read it
Have ten minutes? Read the two-page brief at the front. Evaluating? Read chapters 2 to 6. Asked to defend the choice? Go straight to chapter 10, “Hostile questions, answered”, and chapter 12, “Checklists”. Part two, chapters 14 to 16, shows the work in practice. Ready-made prompts are in the last appendix.
What’s inside
- The memory problem every AI program runs into
- What SAIHM is, in plain words
- Shared memory: a secure, persistent channel between agents
- SHM: what Super-Human Memory adds
- Levers and recipes: tuning SHM for your goals
- For the enterprise
- Between organizations: trade and partnership
- Swarms, fleets and robots
- Why choose SAIHM + SHM
- Hostile questions, answered
- FAQ
- Checklists
- Join SAIHM: the case for now
Part two: the work, in practice
- Scenario: three roles instead of about twenty agents
- Token economy and cost
- Getting started in Claude Code
Appendices cover sources and verification, a glossary, figures and credits, discount codes for developers, other memory layers at a glance, MKT (the next add-on), and prompts.
Prefer to watch?
SAIHM in six minutes: the short video overview, with captions and a full transcript.