Independent AI research from Riyadh
Intelligence should be open, inspectable, and useful.
Najd Research builds open systems for machine intelligence. We are starting with Najd Code—an environment where agents can plan, use tools, delegate work, and finish software.
01 / CURRENT WORK
One active edge.
A connected system next.
Najd Code is the active proving ground. Chat, Craft, Search, Models, and Inference follow from the same open foundation as the next systems to build.
- 01Active development
Najd Code
An environment where agents can plan, use tools, delegate work, and finish software.
- 02Next
Najd Chat
A conversational interface built on a shared knowledge core.
- 03Next
Najd Craft
An agentic builder for applications, documents, and working artifacts.
- 04Next
Najd Search
Retrieval and research across connected, inspectable knowledge.
- 05Next
Najd Models
Models trained, adapted, or released when the work earns that name.
- 06Next
Najd Inference
Open infrastructure for serving, evaluating, and operating models.
02 / PRINCIPLES
The work should explain the lab.
Najd Research exists to make machine intelligence more open, inspectable, and dependable through systems people can actually use.
- 01
Open by default
Publish code, methods, and evidence that make scrutiny possible.
- 02
Systems over demos
Build reliable environments around models, not spectacles around isolated outputs.
- 03
Research through building
Let working software reveal the questions worth answering next.
03 / RESEARCH INDEX
Evidence, when it is ready.
Releases, notes, models, and experiments will appear here when there is something concrete to inspect.
Index status
No public releases yet.
Najd Code is in active development. We will publish the record with the work.