The AI industry has a duty. Here’s one thing we’re doing about it.
Open-source AI memory: pay-as-you-go in pennies, storage-node earning, public-good infrastructure.
AI is changing what people get paid for. SAIHM is one small, real, open way to participate in the change rather than be displaced by it.
Pay-as-you-go: per-call rates with no subscription (current rates on pricing). Use it → · Earn: run a storage node — see setup links.
The premise
AI is automating tasks at speed. The industry that builds AI has a responsibility to provide pathways forward — not because regulators demand it, but because civic stability requires it. SAIHM is a small piece of that. It is open source. It is on a public chain. It does not require permission to use, build on, or operate.
What you can do, with no AI background
- Start with pay-as-you-go. Per-call rates (see pricing) with no subscription and no card-on-file requirement. Give your AI agent a memory that you own. /individuals walks you through it in plain language.
- Run a storage node. The protocol pays storage operators to keep encrypted memory cells available. Anyone can run one — on a home server, a small VPS, or rented capacity. It is one of the few earn-on-AI-infrastructure paths that does not require AI expertise. Setup links: /appendix/storage-nodes.
- Tell a friend. The protocol grows by adoption. If a friend says “I want my AI to remember things, but I don’t want my data on someone else’s server,” tell them: Join SAIHM.
People already doing it
Names invented; circumstances drawn from how the protocol actually fits public-good usage.
- Eleanor, retiree, no AI background. Asks her assistant to remember her medication schedule, her doctors’ names, and which pharmacy stocks her insulin. Pay-as-you-go covers her light use in pennies — no card-on-file requirement, no upsell pressure, no surprise charges.
- Diego, 52, small-business sysadmin. Has unused storage capacity on his off-hours servers. Runs a SAIHM storage node alongside the rest of his infrastructure; the protocol pays storage operators to keep encrypted memory cells available, and Diego can verify what he’s holding (encrypted blobs only, never plaintext) at any time.
- Aliyah, university researcher. Building reproducible AI agent experiments for her PhD. PAYG covers her workload in pennies; no procurement cycle, no time-bombed trial. When a paper goes through review, the audit trail of what state the agent had at each step is independently verifiable on the public chain.
- A small public library. Runs an AI assistant that helps patrons find books, place holds, and navigate library policies. The assistant’s memory of each patron’s reading interests stays patron-local under their own keys; the library cannot read it, and the library cannot be subpoenaed for records it does not hold.
Pricing that scales down to pennies
Pay-as-you-go settles per call at the current rates on pricing. There is no subscription minimum, no card-on-file requirement, and no surprise charges. You pay only for what you write or read.
Pro and above exist for users who genuinely need higher quotas, sharing allowances, and longer audit retention. Public-good access scales naturally at the PAYG floor.
Updates and post-mortems
Material protocol updates — ratified architecture revisions, on-chain anchors, incident write-ups, and governance vote outcomes — are recorded at /blog. The first post lands when there is something material to say. Operational state lives at /status; the dated change history lives at /changelog.
Why we frame this as a public good
Public-good infrastructure has properties: it is non-rivalrous, it is open to all, and it resists capture. SAIHM aims for those properties:
- Apache 2.0 source — nobody can take it private.
- Public-chain anchored — nobody can change the build commitment without the change being publicly visible.
- User-held keys — nobody can take your memory away from you without destroying the key, and only you hold that.