The platform
Learneta
The agentic architectural technology that empowers learning solutions for higher education, the online-programme operating model, and enterprises: the multi-agent system that works with humans to support learning experiences from first touch to outcomes.
Everything Agentic Learning Solutions builds sits on it.
First touch to outcomes
A learning experience is longer than a course.
It starts at a student's first enquiry and runs to the moment the learning has done its work. Along that journey sit dozens of tasks that are repetitive for people and unforgiving of error: the exact territory where a multi-agent system, working with humans rather than instead of them, earns its keep.
The lit stages are where product one operates. The platform is built for the whole line.
Design principles
Two rules run through everything on the platform.
Deliberately non-generative
The AI never writes your content. It makes pedagogical and architectural decisions about how approved content is rendered, sequenced, and delivered. The words stay the words your experts signed off.
Verification you can inspect
Every output carries a computed state: VERIFIED where it traces fully to source, DRAWN where bounded fields come from approved patterns, DRIFT where anything has appeared that the source does not support. The state is computed from the source, not asserted by the model.
Trust in an agentic system is an architectural property, or it is not there at all.
On the platform
Products ship on the platform, one at a time.
Agentic LXD
Focused solely on the LXD process, the design and build of learning experiences. It takes an expert's intent and approved content through to built learning in seconds.
About Agentic LXD →Where this goes
Components that humans and agents can both use.
H5P shipped components that human authors could pick up and use. The next layer is a component library that autonomous agents can use as well: each component addressable as an MCP tool, with a typed schema for the source spans it consumes and the rendering it returns.
Human authors and agent authors call the same tools, and the verification mechanic runs the same way for both. The human-built module and the agent-built one are the same module, built from the same parts.
Status
Built in the open, funded by its own founder.
Learneta is in active build as self-funded R&D, with two working prototypes live on product one and the thinking published as it happens. If the architecture interests you, the writing is where it is worked out.