Nova v2.5: Nova Shell, Nova Transfer, Dashboard Studio

Three big surfaces shipped at once. Conversational infrastructure control. Encrypted cross-cloud file transfer. A drag-and-drop dashboard builder that ships with golden signals pre-wired on every service.

Nova Shell

Nova Shell is a conversational interface for infrastructure operations. Type "scale the payments-api deployment to 8 replicas" and the shell parses the intent, generates the kubectl command, shows you the diff, and waits for confirmation. Type "show me which pods are restarting in production" and it runs the right query and renders the output as a table.

Under the hood, the shell is an LLM-driven planner with a tool palette of 80+ vetted operations, kubectl, terraform plan/apply, AWS CLI, Azure CLI, gcloud, datadog query, and so on. The model picks the right tool, fills in arguments, and emits a structured plan. The plan is shown to the user before execution; nothing runs without explicit confirmation for write operations.

Read operations (queries, listings, log fetches) execute on enter. Write operations (changes, deletions, deploys) require a confirmation click after the user reviews the generated command. This matters, the LLM can hallucinate; the human is the safety net. Every confirmed action is logged in the audit ledger with the natural-language input, the generated command, the diff, and the outcome.

Nova Transfer

Nova Transfer is encrypted, audited cross-cloud file transfer. Move a 50GB dataset from S3 to GCS or from Azure Blob to S3 with end-to-end encryption, RBAC on both source and destination, and a full audit trail. The transfer is chunked, resumable, and bandwidth-throttleable.

The implementation uses streaming encryption, bytes are encrypted client-side before they hit the source's egress, and decrypted only at the destination. Nova never has the unencrypted bytes in memory; we just orchestrate the transfer. The encryption key is derived from a per-transfer ephemeral key plus the customer's KMS root key.

The audit trail is the part most customers asked for. Every transfer logs the source, destination, byte count, hash of the encrypted payload, who initiated it, what RBAC decision was made, and the timestamp. Compliance teams use this to satisfy data-movement audits without instrumenting every file copy manually.

Throughput is the other piece. Naive S3-to-GCS copy through a customer's VPC tops out at 100 MB/s. Nova Transfer pipelines parallel chunks across both source and destination, hitting 1.2 GB/s on standard cloud egress connections. A 50GB transfer drops from 8 minutes to 40 seconds.

Dashboard Studio

Dashboard Studio is the new dashboard builder. Drag a metric onto the canvas; pick a chart type; set the time range. Save. Share. The interaction model is closer to Figma than Grafana, direct manipulation, snap-to-grid, multi-select, copy-paste between dashboards.

Behind the scenes, dashboards are JSON documents stored in a per-tenant database. The editor is a React canvas with a virtualised viewport (so a dashboard with 200 panels still scrolls smoothly). Queries are debounced and deduplicated, if three panels are charting the same metric with the same time range, the underlying request fires once.

The shareable link is the bit that matters most. Paste a Nova dashboard URL into Slack and the unfurl renders the live current state of the dashboard, not a snapshot. Click through and you land in the read-only viewer with the same time range that was active when the link was shared.

Golden signals on every service

The bigger Dashboard Studio change is what ships out of the box. Every service Nova discovers automatically gets a dashboard with the four golden signals, latency, traffic, errors, saturation, pre-wired with sensible chart types and time ranges. The dashboards are auto-generated; you don't pick the metrics.

The auto-generation pulls from the service's instrumentation. If the service emits Prometheus metrics with the standard names (http_request_duration_seconds, http_requests_total, etc.), the dashboard wires up correctly on first detection. If the service uses non-standard metric names, Nova suggests the closest matches and the user confirms or remaps.

The result: an engineer onboarding a new service sees a working dashboard within minutes of deployment. No manual dashboard authoring; no Grafana JSON wrangling; no copy-paste from a template that's three months stale. The dashboard is a derivative of the running service, not a separate artifact to maintain.

Why this all shipped together

The three features look unrelated. They're not. Nova Shell, Nova Transfer, and Dashboard Studio share one underlying primitive, the audit ledger. Every action across all three is logged the same way, with the same structured fields, queryable through the same API.

That uniformity is what makes the platform feel coherent. The on-call engineer who used Nova Shell to scale a deployment can answer "what changed in the last hour?" by querying the same ledger that captures dashboard edits and file transfers. The compliance officer can satisfy "show me every change to production this quarter" with one query.

v2.5 is the release where that ledger went from internal plumbing to a customer-facing capability. Future releases will lean on it more, that's the architectural bet.

All three features are live in production for all tenants today. Nova Shell is in the global header; Nova Transfer is under the Operations tab; Dashboard Studio replaces the previous dashboard editor with a one-click migration of existing dashboards.