Our work · Status: In development
One pebble. A wider learning ecosystem.
Grounded learning story
The Pebble work ecosystem
- 01Question
A learner or responsible adult starts with a defined learning need.
- 02Source
The system identifies relevant, approved educational knowledge.
- 03Retrieve
Only a small, useful context is selected for the task.
- 04Explain
Pebble prepares an explanation or draft and makes its limits visible.
- 05Review
A person checks the output and decides what happens next.
This is an intended response pattern for a system in development. Retrieval can improve relevance; it does not make AI infallible.
Capability with a purpose.
Pebble is the shared centre: a free educational AI initiative in development. The surrounding work explores what it needs to become useful, accountable and available in real learning environments.
Pebble Knowledge
In development · Curated, traceable educational sources for grounded answers.
Pebble Models
Research direction, not a completed finding · Exploration of language and learning models suited to local constraints.
Pebble Bench
Research direction, not a completed finding · Evaluation for usefulness, safety, language quality and groundedness.
Pebble Classroom
In development · Role-aware learning support shaped with educators and responsible adults.
Pebble Local
Research direction, not a completed finding · Local-first infrastructure for places where connectivity is limited or costly.
Open work
Open work · Methods, documentation and selected tools shared when useful and responsible.
Build the foundations before the promises.
Current work centres on defining safe use cases, assembling appropriate knowledge, testing retrieval and evaluation approaches, and learning from the adults responsible for education and care.
No initiative listed here should be read as generally available unless its page explicitly says so.
Have a useful problem to explore?
Tell us about the learning context, constraints and people who would remain responsible for the outcome.