SLM as Totebox sysadmin and support centre
docs(2j): vocab scrub — services/ + systems/ (9 files EN+ES)
@@ -95,7 +95,7 @@ The service-slm-as-sysadmin capability is available across three customer tiers, **Tier 2 — Multi-Totebox customer.** Tier 1 plus access to vendor-hosted Tier B (OLMo 3.1 32B Think) for harder queries. Per-tenant LoRA adapters trained on the customer's own corpus deploy on Tier B. Pricing: per-token at PointSav rates, planned to be structurally below the $95,000–$590,000 annual floors that gate comparable closed-source SMB offerings. *Planned pricing is intended to be structurally favourable relative to those alternatives; actual pricing is not yet set and is subject to operator decision before any commercial offer.* [ni-51-102] [osc-sn-51-721] **Tier 3 — PointSav-LLM specialist product.** A vendor-hosted model trained on Foundry's aggregated multi-tenant corpus plus curated public corpus; a specialist in Totebox Archive operation, PointSav conventions, and the ten task families. L1 AI resolution of a planned 80–90 percent of customer queries autonomously; L2 human-in-the-loop escalation; L3 engineering for rare complex cases. Each L2 response feeds back as DPO training signal. *Tier 3 deployment timeline is planned for v0.5.0 (Q1 2027 per current roadmap); actual timing is subject to corpus accumulation rate and operator decision. [ni-51-102] [osc-sn-51-721]* **Tier 3 — PointSav-LLM specialist product.** A vendor-hosted model trained on the platform's aggregated multi-tenant corpus plus curated public corpus; a specialist in Totebox Archive operation, PointSav conventions, and the ten task families. L1 AI resolution of a planned 80–90 percent of customer queries autonomously; L2 human-in-the-loop escalation; L3 engineering for rare complex cases. Each L2 response feeds back as DPO training signal. *Tier 3 deployment timeline is planned for v0.5.0 (Q1 2027 per current roadmap); actual timing is subject to corpus accumulation rate and operator decision. [ni-51-102] [osc-sn-51-721]* ## See also