Point your OpenTelemetry collector to NemiOS. Zero code re-instrumentation. In seconds you see metrics, logs and traces populating the 8 senses. If you already have OTel partially deployed, migrating is changing the endpoint. If you use proprietary SDKs, we include an adapter layer for syslog, statsd and Prometheus scrape.

Nothing rebuilds. Dashboards, alerts, runbooks, integrations, historical data and Copilot's accumulated context travel as-is. The only thing that changes are the limits — AI queries, retention, active integrations, SLA, data residency. It's not migration; it's a flag that activates.

During onboarding we capture your business profile with 3-4 short questions: industry (free text), size, critical services and seasonality. A parametric engine combines annual revenue, service share, peak hour and degradation severity. Every figure is expandable to its reasoning chain with inputs you can edit. Seven validation layers before showing the number.

Chispa, Órbita and Galaxia run in the closest region with good latency from day one. Mexican data residency (standard option in Cosmos, on AWS MX) arrives with Phase 2. Nemi Private — full deploy inside the customer's perimeter, on-premise or private VPC, without telemetry crossing to NemiOS — also arrives in Phase 2 (see ADR-0006).

Never, without explicit human approval. From day one, Copilot operates in suggestion mode: proposes the next action, you sign with one click, the system executes it and the audit trail is captured. Autonomous mode — for reversible, low-risk actions (pod restart, scaling, rollback, feature flag) — arrives with Phase 2, with instant kill-switch and complete audit log. Every action has signature, timestamp and justification.

No invoice overage. The deterministic router automatically degrades to the cheapest model of the next tier (Opus → Sonnet → Haiku) on the service that exceeded, and notifies you via WhatsApp with the option to upgrade commercial tier. Never an invoice surprise.

Tier-1 humans in Mexico City; tier-2 and tier-3 distributed across LATAM with 24×7 coverage on Cosmos. WhatsApp Business as the official first-class channel — alerts, natural-language queries and approvals. Same-day response native to your region — never 12-hour delays lost across time zones.

From day one, Audit gives you the immutable audit log of every action, approval and change with timestamp — sufficient base for internal audit and risk committee. Specific regulatory reports for CNBV, Circular Única de Bancos, Ley Fintech and LFPDPPP arrive with Phase 2, with a dedicated agent that reads new circulars and proposes automatic observability adjustments.

Initial time-to-value measured in days, not weeks. Guided onboarding of 3-4 questions + OTel auto-instrumentation + generative dashboards from the first moment. The first incidents translated to money appear the same day you connect your telemetry.

NemiOS automatically redacts LATAM-regulated patterns at the OTel Collector at ingress time — before the data touches storage or leaves your perimeter. Built-in patterns: CURP, RFC, CLABE, PAN with Luhn validation, CVV, emails, MX/CO/BR/AR/CL phone numbers, IBAN, Brazilian CPF, Argentine DNI, Chilean RUT and postal addresses. You can add custom regex for internal IDs. The Collector configuration is auditable by your security team before production — verifiable technical control, not a contractual promise.

Every action executed by Copilot, every approval, every config change, every inbound webhook — all logged in an immutable audit log stored in ClickHouse with 5-year retention. Append-only, with envelope encryption of third-party credentials, HMAC-SHA256 on inbound webhooks, and structural separation between who proposes (the AI) and who executes (the integrations service). Auditable per tenant, exportable anytime for risk committee or external auditor.

Yes. NemiOS exposes a public API with OpenAPI 3.1 from the first release. The spec published at docs.nemios.ai is the source of truth — official SDKs are generated from there. URL versioning /v1/, API keys with prefix nk_live_ and nk_test_, RFC 7807 errors with trace_id, rate limits per tier. Contract-first: spec changes go before code, and CI detects breaking changes automatically before any deploy.