Technical deep dives, product updates, and practical engineering insights for Data + AI teams
MOI 5.0 is in public beta. This Saturday, Genesis offers six more hours of free GLM 5.3 and Kimi K3 access with no token cap, plus 70% off prepaid credits.


MOI 5.0 public beta is now live for all users. On November 5, Genesis offers GLM 5.3 and Kimi K3 free for six hours, and four perks — activation, feedback, tasks, and referrals — remain open.

The hidden systems problem behind long-context AI: giving an agent more material is easy; getting it the right material at the right moment is the hard part.

MOI developer mode is live!From a local terminal, developers can ask an agent to query data, import and export files, run existing workflows, create knowledge bases, and add content to them on MOI.

MatrixOrigin officially launches Genesis Token Service, an integral part of MatrixOne Intelligence 5.0. Scenario-aware, data-aware, and cost-aware, Genesis optimizes token efficiency from applications to infrastructure, making model services more efficient, economical, and transparent.

MatrixOne Intelligence 5.0 is officially opening its second round of public beta, featuring new direct document and email ingestion capabilities — come try it out with your real business needs!

As models, compute, and tokens become easier to acquire, what can enterprises truly own and compound? The answer begins with proprietary data—and how it is connected to real business execution. From data and tokens to business-system reinvention: rethinking the long-term value of enterprise AI.

When AI moves from generating answers to carrying business tasks, the hard problem is state, not the model. This white paper presents AI State lifecycle management, the six components of the MOI platform, and measured results from four customer scenarios.

GitHub is rebuilding Git for agent-scale development. Data infrastructure may be next.

MatrixOrigin hosted an AI Leadership Dinner in Silicon Valley, welcoming a delegation of Chinese AI industry companies visiting NVIDIA headquarters and bringing industry partners together to explore enterprise AI adoption and collaboration.

Claude Code Projects and procedural memory are pushing agent experience closer to execution. This article examines why shared, writable memory needs versioning, provenance, conflict handling, and rollback.

This article uses an Agent Memory experiment to examine the accuracy, cost, and failure modes of a JEV judgment layer, and explores when adding a helper model to an Agent is actually worthwhile.

Enterprise AI must connect two worlds: deterministic software that executes reliably and probabilistic agents that reason, use context and learn from feedback.