One pane of glass for every Avyra tenant.
Avyra Dashboard is where operators actually run the business. Calls, intents, latencies, occupancy, revenue, errors — every number the suite emits, in a layout designed for shift handoffs and quarterly reviews, not screenshots in a pitch deck.
What it does in production.
Live call & order analytics
Volume, duration, intent breakdown, sales by hour — drilldowns by tenant, station, or staff member.
Latency histograms
STT / TTS / LLM percentile latencies, surfaced where it matters. Spot model degradation before users do.
Tenant management
Add a clinic or venue, configure prompts and FAQs, manage staff users — all without touching the YAML.
Alerts & on-call
Wire to PagerDuty, Slack, or email. Latency spikes, queue depth, missed-call rates — pick what wakes you.
Call & transcript search
Full-text search across every captured call. Replay the audio, read the transcript, jump to the intent timeline.
Suite-wide overview
Dine, POS, Voice, Vision — one home screen, every product, every venue, no tab-juggling.
Prompt & FAQ editor
Update the receptionist's prompts or the clinic FAQ from the UI. Changes pick up without a redeploy.
Audit log
Every config change, prompt edit, and tenant action — written to an immutable log with actor and timestamp.
How it's built.
- Dark-first design system shared with the rest of the suite (violet accent over near-black).
- Reads Prometheus for latency / volume metrics, Postgres for call & order history.
- Tenant boundary enforced at the gateway — no cross-tenant leaks at the data layer.
- Live updates over WebSocket; falls back to polling if WS isn't available.
- Designed mobile-aware so a manager can spot-check the shift from a phone.
- Ships as a single Dockerized service with nginx in front.
┌──────────── Dashboard (React) ───────────┐
│ Inter / JetBrains Mono · dark-first │
│ Multi-tenant routing & RBAC │
└────────────────┬─────────────────────────┘
│ REST + WebSocket
▼
┌─── tenant-aware API gateway ────┐
│ │
▼ ▼
Postgres (logs, configs) Prometheus (metrics)
▲ ▲
│ │
└── ai-engine, POS, Dine ──────────┘
(each emits both)Who runs it today.
Owner / operator
Daily revenue, top items, missed calls, busy hour — everything you'd ask the manager, on one screen.
Engineering
Latency dashboards, error budgets, prompt experimentation. Where the team watches a model rollout.
Multi-tenant SaaS
If you operate Avyra for multiple clinics or restaurants, the dashboard is your operator console — one login, every tenant.
Questions we get asked.
Is this Grafana with extra steps?+
No. Grafana is great for raw metrics. Dashboard is a product-aware operator console — it knows what a 'missed-call' is, what 'shrinkage' means, and how to surface them to non-engineers.
Can I host it separately from the rest of the suite?+
Yes. It only needs read access to the API gateway and Prometheus. Run it in its own VPC if you want.
Mobile support?+
Reads well on phone, but it's designed for a desktop monitor. We don't pretend it's a phone app.
Got a product to build? Tell us what you have in mind.
We kick off in days, not months. Working software in weeks. If we're not the right fit, we'll tell you up front.