How AvioBook turned flightroom data into plain‑language turnaround insights with Amazon Bedrock AgentCore

# The operational problem Airlines operate on tight, cascading schedules where minutes at the gate compound across a day. AvioBook captures every turnaround as a flightroom: a durable archive of timestamped API events (aircraft changes, delays, boarding progress) plus the conversational messages crews and operations exchange. That archive contains the data needed to explain delays, but it usually requires time, queries, or a dedicated analytics team to surface answers.

# What AvioBook built

Two distinct agents were created to match the different needs of the two roles: one focused on historical pattern analysis for managers, the other on immediate operational verification for dispatchers.

# Why this matters to airlines Collecting data and using data are different problems. AvioBook identified three recurring operational gaps:

• Operations were effectively a black box: teams on the same turnaround lacked a consistent, real-time picture and accounts differed during audits.
• Delay codes told half the story: coding is done under time pressure and often reflects the dominant delay rather than the upstream cause.
• Answers required a data team: historical questions usually meant manual queries or extraction requests many airlines can't afford.

Connected Analytics aims to reduce reliance on manual review and specialist analysts by putting grounded answers where people need them: in flightrooms and in the hands of managers.

# Architecture and design choices AvioBook used Amazon Bedrock AgentCore to prototype a multi‑agent proof of concept. Key design elements in the architecture include:

• Persona‑specific agents: separate agents for managers and dispatchers so outputs are tailored to each role's timeframe and evidence needs.

• Evidence delivery: answers are returned with the specific events and messages that support the conclusion, making them usable for audits and compliance.

# Practical results reported AvioBook shared that the Connected Analytics prototype turns flightroom archives into plain‑language, evidence‑based answers for operational teams. Customers using AvioBook Connect have begun seeing those turnaround data used in the moment rather than only for post‑hoc review. The architecture is positioned to reduce time spent cross‑referencing histories and to help surface upstream causes that delay codes can miss.

# What this implies for operations teams Using agents to extract and summarize flightroom data removes a layer of manual work and supports quicker, auditable explanations of what happened during a turnaround. For mid‑size carriers where turnaround is a binding constraint, reducing the time to insight and the ambiguity around delay causes can change how on‑time performance issues are assigned and remediated.

# Bottom line