How it works
The five-layer YAAIF stack: Channels, Agent, Skills, Capabilities, and LLM Models. Every request travels the same governed path through these layers, with its own audit trail at each step.
The platform
How the YAAIF enterprise agentic AI OS is built. Five architectural layers, three agent patterns, configurable channels, three deployment options, an end-to-end trust posture, and the Third Eye observability layer that makes every action reconstructable. Open standards, no vendor lock-in, compliance built in from day one.
Six platform topics
The platform is one stack with one governance posture. The pages below let you enter where your team's decision actually sits, whether that is architecture, integration, or audit.
The five-layer YAAIF stack: Channels, Agent, Skills, Capabilities, and LLM Models. Every request travels the same governed path through these layers, with its own audit trail at each step.
Three agent patterns covering nearly every operational workflow: Chat Agent (natural-language with structured input cards), Desktop Agent (Computer Use on legacy apps), and Ambient Agent (always-on event response). All three share the same governance.
Where work comes in: Microsoft Teams, Slack, Google Chat, email and ticket queues, voice and telephony, APIs and webhooks, event streams, schedules, plus MCP hosts (Claude Desktop, ChatGPT desktop, any MCP-compatible client).
Three deployment models with the same platform and governance: bare-metal Linux (including air-gapped), your cloud tenant on AWS/Azure/GCP/Kubernetes, or a managed SaaS environment we operate for you.
The control posture that holds up to internal audit and regulators: governance by default, federated identity, run-level audit, data residency, ownership of configurations, and versioned change control.
The Third Eye: every request correlated across seven layers (Channel, Agent, Skill, Tool, Model, Memory, Response). Cost attribution per user, agent, and workflow. Audit kept separate from telemetry. Drops into OpenTelemetry, Prometheus, and your existing log stack.
The same governance posture applies in every layer. Policy gates, human approvals, identity boundaries, and an append-only audit trail apply on every channel, in every pattern, in every deployment model.
Next step
A 30-minute working session walks through the platform layer your team is most concerned about, from integration surface to audit trail.