Tiered inference gateway — local-first AI routing
editorial(substrate): fix soft-slm-tiered-gateway's GPU model + MCP transport claims (Track-B) — confirmed express-node GPU is L4 (opentofu main.tf), 'A100' is stale internal code-comment language, same finding already established on ai/zero-container-inference.md; confirmed the MCP interface is a separate stdio-transport binary (slm-mcp-server) speaking HTTP to the Doorman as its own client, not a second network port the Doorman itself exposes; fixed pre-existing duplicate body H1 (content-contract violation) EN+ES; register-clean EN+ES
@@ -18,8 +18,6 @@ paired_with: soft-slm-tiered-gateway.es.md supersedes: slm-tiered-substrate.md --- # The tiered inference gateway — local-first AI routing A tiered inference gateway routes every AI request through a hierarchy of compute tiers, selecting the least expensive capable tier for each request. Routine work runs on hardware the organization owns. Burst capacity on rented GPU handles work that @@ -118,8 +116,8 @@ dispatching to that tier immediately. In-flight requests complete; no new reques start. Queued work accumulates and drains when the kill switch is reopened. The kill switch is the operator's billing control. Closing the express node switch stops the A100 from starting; the cost drops to zero. Closing the global switch stops all Tier B and Tier C spending while allowing Tier A to continue serving. stops the L4 GPU node from starting; the cost drops to zero. Closing the global switch stops all Tier B and Tier C spending while allowing Tier A to continue serving. The express lane — which bypasses the file-backed queue for time-sensitive work — still checks the kill switch. Nothing bypasses the kill switch. @@ -152,11 +150,14 @@ slow graph service from blocking inference. ## The MCP server The gateway exposes an organizational memory interface via the [[mcp-substrate-protocol|Model Context Protocol]] at a second port. Any MCP-capable AI client can connect to this interface using its built-in subscription — no separate API key is required. The client's reasoning capability combines with the gateway's [[ontological-datagraph|organizational knowledge graph]] to produce responses that are grounded in the organization's actual data. A companion process exposes the gateway's organizational memory as [[mcp-substrate-protocol|Model Context Protocol]] tools — not a second network port on the gateway itself. It is a separate program, speaking MCP over standard input and output to whatever client launches it, and speaking ordinary HTTP to the gateway as its own client. Any MCP-capable AI client can launch this process using its built-in subscription — no separate API key is required. The client's reasoning capability combines with the gateway's [[ontological-datagraph|organizational knowledge graph]] to produce responses that are grounded in the organization's actual data. This is the primary path for interactive use by operators who already have a subscription to an MCP-capable client. The gateway handles memory; the client