The most important August decision was a separation of responsibilities. Grace is the operating and orchestration layer controlling the full system. R.A.I.S.E. is the Redacted AI Secure Environment, the protected kernel and runtime where local retrieval, models, tools, and agents execute.
Why the distinction matters
Grace and R.A.I.S.E. had sometimes been described too close together. That made it hard to explain what governed the whole system and what secured a particular execution environment.
The new architecture made the boundary explicit. Grace owns orchestration, identity, knowledge, policy, tools, agents, workflow state, approvals, and decision history. R.A.I.S.E. provides protected local retrieval, inference, tool execution, evidence handling, and agent runtime.
The model moved down the stack
A model is an engine selected for a task. It should not own organizational memory or force the company to rebuild its workflows around a vendor. Grace routes work. R.A.I.S.E. provides the environment. The model performs bounded inference inside that structure.
The architecture can support any practical model class when licensing permits and the selected infrastructure provides enough memory, compute, storage, power, cooling, resilience, and budget. That describes capacity. It does not mean the portable Z13 runs any model you name.
The vertical becomes a deployment pack
Research, legal, government, medical, and enterprise systems should not turn into unrelated product lines with separate cores. The OS, secure runtime, and governance fabric remain consistent. The corpus, professional rules, workflows, approvals, evaluation, and hardware profile change by domain.
The public focus became research, edge technology, and legal systems as core specialties. Medical and health research, government and public sector, and enterprise private RAG remained qualified opportunity lanes rather than unsupported claims of universal readiness.
The website became an architecture map
The August rebuild reorganized the homepage, navigation, platform page, metadata, social card, sitemap, pilot, and vertical pages around the product stack. Grace sits above governance. R.A.I.S.E. sits above data, models, hardware, and the network boundary.
The rebuild created a stable public language the company can use in technical discovery, partner conversations, pilots, and future deployment documentation.