what it enables
Use cases that do not start with a silo
sempods is young, but the pattern is already visible: when data has its own address, meaning and permission boundary, applications can cooperate without one of them becoming the owner.
Events as Linked Open Data
This is the most concrete example today. Venues manage events in an app, but the public event data is served from their own pods as JSON-LD and can be queried without an API key.
A venue website can read the same data. An aggregator can combine multiple pods. An AI agent can answer what is happening this week. Nobody has to own a central event database for the model to work.
Try the aaltra pod on the Start page. It is live data from a real venue in Chemnitz.
AI memory under your control
Today, every assistant wants its own memory. That is convenient until you switch tools, change providers, or ask what the assistant actually knows about you.
With sempods, memory can be structured data in a pod. An AI agent receives access like any other client: read these contexts, write to that context, lose access when the token is revoked.
The important part is not that AI is special. It is that AI is finally a client powerful enough to use a personal knowledge graph well.
Knowledge management for products and teams
Many companies are trying to make AI answer from internal knowledge without copying everything into yet another proprietary system. The hard part is not only retrieval. It is structure, permission and provenance.
sempods fits that shape: knowledge can stay distributed, permissions remain explicit, and apps or agents can query the graph they are allowed to see.
This is where adoption by independent players could matter most: the personal data vision becomes stronger when organizations use the same open model for practical reasons.
Open knowledge publishing
Museums, cities, research groups and communities often have knowledge that should be linkable and reusable, but not locked into one website or export format.
A pod can publish public contexts as Linked Open Data and keep other contexts private or paid. The public layer gives the web something to link to. The private layer keeps curation, workflow and access control with the publisher.
What is still missing
The foundation exists, but the easy paths are still being built. The public TypeScript SDK, the dedicated specification repository and betterfind adapters are the next pieces that make these use cases easier to implement outside the reference stack. SHACL shapes and a TBox model remain important design work, but they are not final public contracts yet.