Luma

A RAG-based AI agent for airline knowledge work.

Luma answers from your governed knowledge base, with citations. It does not pretend to be a customer-service replacement or a system that can take destructive actions.

What is the cutover sequence for a Crossing migration?
Architecture → data-migration design → parallel run → cutover governance → post-cutover stabilization. Delivered in alignment with Sabre. Want the typical timeline?
Yes, and the main risk.
Six to eighteen months. The principal risk is parallel-run data drift, mitigated by reconciliation gates before cutover.
Operator-Built

Built by an operator who has shipped production airline AI.

Mike conceived and launched Sabre’s first production agentic AI for airlines, so Luma is built with hard-won judgment about what production airline AI actually requires. Luma is Alegris’s own retrieval-grounded agent: built independently, citation-backed, and configured against a corpus you control. For this site, this site, Luma’s corpus is the Alegris catalog itself.

Try It

Ask Luma about Alegris.

This demo is grounded in the public Alegris content: the practices, the services, and the journal. Every answer comes back with a citation.

LumaRAG · grounded in Alegris content

Hi, I'm Luma. Ask me anything about Alegris: the practices, the services, or how the firm works. I answer only from the Alegris knowledge base, with a citation.

Three Use Cases

Where Luma earns its keep.

Airline knowledge base

Operations manuals, fare rules, platform documentation, answered conversationally with the source cited, instead of buried in a portal.

Training tutor

A self-paced tutor for Radixx Res, Galaxy, and the API surface. This is the line-agent layer of Radixx Academy, available on demand.

Conversational analytics

Ask your commercial and operational data questions in plain language. This is the layer that powers Radixx Lens.

Deployment Posture

Aviation-grade by default.

Luma is deployed with a compliance posture aligned to GDPR, PII, and PCI requirements, with the knowledge boundary and retention rules set per engagement. We are explicit, up front, about what is in scope and what is not.

What Luma is not: not a customer-service replacement, not a marketing chatbot, and not an agentic system that can take destructive actions. The boutique reads as serious by being honest about limits.

Talk to a principal about deploying Luma.

We will configure Luma against a real knowledge base and show you the lift, honestly.

Talk to a principal →