PROS helps airlines make faster, data-driven decisions about the offers they present to travelers. As airline retailing moves toward real-time, personalized offers, PROS needed the right partner to deliver intelligence from day one in its new PROS Kairos platform. With the Databricks Data + AI Platform, the team grounded Genie Agents in its airline data, governed model calls through Unity Gateway and adopted Lakebase Postgres. Questions that once took days now take minutes.
A wealth of airline data requires faster access for users
For more than 40 years, PROS has helped airlines forecast demand, optimize pricing and improve commercial performance. Today, that expertise is powering the next generation of AI-driven airline retailing.
“We don’t see AI as just another feature. It’s fundamental to how the next generation of airline retailing will work, and if we don’t build it, somebody else will,” said Hari Iyer, Software Architect at PROS.
Business users relied on predefined reports, and any new question became a ticket routed to engineering, so answers typically took a couple of days. An earlier version of AI services at PROS let a large language model (LLM) query the data warehouse directly. Still, without curated context, answers could vary from one request to the next.
“We know how to build an AI platform, but then we’d have to maintain it, and that’s not the business we’re in,” said Daniel Santana, Software Architect at PROS.
Integrating Genie Agents, Unity Gateway and Lakebase on Databricks
After adopting Databricks as its analytics foundation, PROS focused on making domain-specific, governed AI available to internal teams and customer-facing experiences. That work now powers PROS Kairos, the agentic airline retailing platform PROS launched in October 2026 with the Databricks Data + AI Platform as its core data and AI foundation.
Building domain-specific Genie Agents for airline retailing
The team built domain-specific Genie Agents for demand, pricing and flight analysis. Each is grounded in curated datasets and internal documents that explain airline terms. When a user asks a question in PROS Kairos, a central orchestrator running as a Databricks App routes the request to the right Genie Agent, MCP server or skill.
“Genie sits right next to the data, set up with sample questions and instructions, so our models simply hand those questions to it. We get faster, stable answers, again and again,” said Murat Berk, Senior Architect at PROS.
Centralized AI governance with Unity Gateway, plus modernization with Lakebase Postgres
Every agent at PROS reaches its models through central endpoints on Unity Gateway. The team picks the right model for each task, logs every request, blocks sensitive information, caps usage and falls back to another model if one fails. Custom endpoints with their own guardrails and permissions support AI management requirements at PROS as it pursues ISO 42001 certification. A natural-language flight search for travelers, now in beta with five airlines, is moving onto the gateway too.
“Having every provider, every model and our own guardrails in the same platform makes it so much easier to control how our agents behave,” Daniel said.
Some PROS travel products rely on large databases sized for six to eight hours of heavy nightly processing and largely idle otherwise. Proofs of concept showed even small Databricks compute outperforming those databases, so PROS committed to Lakebase Postgres as its new transactional backend. In 2027, a partner will migrate 75–100 database instances, including those on Oracle, to Lakebase Postgres so compute scales up for heavy processing and back down afterward. Lakebase Postgres already stores session data for AI agents at PROS, and new applications now default to it.
Faster airline insights and new capabilities through Databricks
With analytics, AI governance and data storage unified on Databricks, insights now reach decision-makers far faster:
- Instant business self-service: Analysts who once waited days for custom engineering reports now ask Genie Agents directly and get answers in minutes, with room for follow-up questions in the same flow.
- Faster time to market: New data-dependent features move from concept to launch substantially faster because datasets and instructions are already accessible through Genie Agents.
- Saved engineering overhead: Unity Gateway eliminated the need for a custom AI gateway, freeing up months of platform engineering time to focus on new PROS Kairos features.
- Accelerated developer productivity: Engineers use Genie Code to prototype agents, draft notebooks and generate dashboards, and it explains the code as they go.
“It’s almost incalculable how much time Genie saved us in getting from where we were to where we are now,” Daniel said.
The Databricks account team worked closely with PROS to accelerate development of PROS Kairos. “Our solutions architect, Thinh Nguyen, knows a lot, and the support has been incredible, even on weekends. It’s the best service I’ve experienced in more than 25 years of working with software,” Daniel said.
PROS plans to extend Genie Agents across capabilities within the PROS Kairos platform covering all four pillars of Product Management, Offer Construction, Offer Optimization and Offer Marketing, while adopting real-time compute for Genie and smart routing in Unity Gateway. With one central orchestrator ready to take on new agents and skills, each new capability reaches airlines sooner.
PROS on Databricks Genie Agents, Unity Gateway and Lakebase
Genie Agents are domain-specific agents on the Databricks Platform that let teams ask natural-language questions about enterprise data. PROS grounds each agent in curated datasets and internal documentation, with sample questions and instructions to ensure consistent responses.