Skip to main content

Frequently Asked Questions

Quick answers to the questions we hear most often. For deeper topics, follow the links into the guides.

Platform​

What problem does Nexa solve?​

When a flight is cancelled, delayed, or rescheduled, the airline typically has to accommodate stranded passengers in hotels overnight. The traditional workflow — spreadsheets, shared inboxes, phone calls, manual bookings — is slow, error-prone, and expensive. Nexa replaces that with a policy-driven, fully automated pipeline: event in → voucher out, in under two minutes.

Who uses Nexa?​

  • Airline operations centers monitoring disruptions and managing bulk accommodation.
  • Airport-based operators handling local passenger reaccommodation.
  • Finance teams reconciling hotel spend and tracking savings from direct contracts.
  • Passengers receiving vouchers and self-serve rebooking information.

Is Nexa a SaaS or a self-hosted platform?​

Nexa is a fully-managed multi-tenant SaaS. There is nothing to install, deploy, or operate on the airline's side — integrations are HTTPS calls to a regional endpoint (for example https://us-central1.api.nexastudio.io) and human users work in the operations console at ops.nexa.ai. See API Access.

What's Nexa built on?​

A managed cloud platform on Google. Customers consume Nexa through the public API and the operations console — there is no stack to install, monitor, or scale. The intelligence layer is a set of Nexa Trained Models on Google powering the flight-disruption predictor and the AI agents. External systems (Amadeus, Hotelbeds, SendGrid, payment) sit behind swappable adapters maintained by the Nexa team, so providers can be added or replaced without changing your integration.

Integrations​

Which hotel providers does Nexa support?​

Out of the box: Amadeus (Self-Service API) and Hotelbeds (APITUDE). Both participate in search and booking. Direct hotel contracts are a first-class third provider — see Contracts.

Can I plug in a different GDS or booking engine?​

The provider interface is small (search, book, optionally cancel). New adapters are added by the Nexa team based on customer demand — talk to sales@nexa.ai if your stack uses a provider we don't support yet.

What about email, SMS, or voice notifications?​

Email via SendGrid is the default. SMS via Twilio and voice via a call-center integration are available on request and configured per tenant — there is nothing to install on your side.

Does Nexa need internet access to every provider?​

Nexa calls providers from its own infrastructure. Sandbox tenants run against limited sandbox providers, so a sandbox can be exercised end-to-end without provisioning a single live credential.

AI and models​

What AI models does Nexa use?​

Four Nexa Trained Models on Google, each scoped narrowly:

  1. A trained flight-disruption predictor.
  2. A deterministic allocation scorer tuned per policy.
  3. A policy synthesizer that converts English into structured policies.
  4. An exception agent that triages manual-review items.

See AI & Models Overview.

Is the AI making bookings on its own?​

No. The exception agent operates via an allow-list of read-only tools. It can search, simulate, and classify — but it cannot book, cancel, or notify. Every action is human-approved. This is enforced at the adapter layer, not just the prompt.

Can I turn AI off?​

Yes. Every AI capability is a per-tenant configuration toggle in the operations console:

  • Exception agent — on / off
  • Policy synthesizer — on / off
  • Flight predictor — on / off

The core pipeline (policy → case → demand → allocation → booking → notification) is fully deterministic and runs end-to-end without any AI agents or trained models.

What data does the AI see?​

  • Flight predictor — weather, traffic, labor, operational signals. No passenger data.
  • Policy synthesizer — the operator's English description + the active policy schema. No case or passenger data.
  • Exception agent — the case, allocation wave, booking attempts, and policy. Passenger identifiers are reference IDs, never passport numbers or full names.

Security and compliance​

How is passenger data handled?​

Nexa stores reference IDs, not documents. Full passenger records (passport, national ID) live in the airline's system-of-record. Nexa sees a pax reference, type (ADT/CHD/INF), and any SSR flags the airline shares.

Are credit cards stored in Nexa?​

No. Card data is never persisted. Payment tokenization is handled by a PSP adapter; Nexa only stores the token. See Payments & PSP.

Is audit complete?​

Yes — every state change writes an append-only audit entry with actor, reason, and before/after snapshots. See Audit.

Is Nexa multi-tenant?​

Yes. Tenant isolation is enforced at the ORM layer — cross-tenant queries return 404 regardless of how the request is constructed. See Multi-tenant Setup.

What compliance frameworks align?​

SOC 2 (Type II), GDPR, and PCI-DSS SAQ-A (thanks to the tokenized payment flow), plus the common aviation-specific audit expectations. Contact sales@nexa.ai for the compliance packet, DPA, and current attestation reports.

Operations​

How is Nexa scaled for a major disruption?​

Nexa autoscales horizontally based on queue depth. Each stage (allocation, booking, notifications) scales independently — provider rate limits, not Nexa capacity, are the practical ceiling on throughput. Customers don't provision or tune anything.

How fast is "event to voucher"?​

Under two minutes end-to-end on the happy path with sandbox providers. With real providers, the bottleneck is Amadeus/Hotelbeds latency — typically 5–30 seconds per booking call — so a full case resolves in 2–4 minutes. Pre-warming via the flight predictor shortens this further.

What's the expected uptime?​

Nexa targets 99.9% on the control plane and publishes live availability at status.nexa.ai. The workers and queues are designed to tolerate provider blips with bounded retry and dead-letter semantics.

What about regional residency?​

The API is deployed in multiple regions (us-central1, europe-west1, southamerica-east1). Tenant data is pinned to a single region for residency; multi-region routing is available for global airlines. Contact sales for the residency you need.

Pricing​

How is Nexa priced?​

Primarily per-disrupted-passenger (with committed-volume tiers) plus a platform subscription. Direct-contract savings are typically 10–20% on hotel spend, which usually covers the platform cost in the first quarter. Custom pricing for large airlines — sales@nexa.ai.

Is there a free trial?​

A sandbox tenant with limited sandbox providers is available for evaluation. Contact sales for a trial of the live pipeline.

Getting help​

Was this helpful?