IsraelTechnology and telecom2025 – ongoing

Wix.com

AirBot, a Slack AI agent for data pipeline failures

Wix built a Slack agent that investigates data pipeline failures and estimates it saves 675 engineer-hours a month

WorkingCompany's own claimOfficial ↗

Problem

Wix runs more than 3,500 Apache Airflow pipelines. When an alert fired, engineers spent about 45 minutes on a typical manual debugging cycle.

Approach

Wix's data engineers built AirBot, an AI agent in Slack. On an alert, it pulls log errors, queries tools such as Trino and Spark, checks table schemas and lineage, and names a likely root cause. It can open a GitHub pull request with a proposed fix and route the alert to the owning team. It supported 30 team channels and 60 data engineers over the 30 days measured.

Results

675 hours a monthengineering time saved, Wix's estimate, about 4 full-time engineersCompany's own claim[1]

28 of 180candidate pull requests merged with no human code changes (15%), per WixCompany's own claim[1]

66%share of 4,200 successful flows that received positive feedback, per WixCompany's own claim[1]

Takeaway

The 675-hour figure is an estimate: Wix multiplied about 2,700 helpful interventions by an assumed 15 minutes saved each. Only 15% of candidate fixes were merged without human edits.

Sources

When AI Becomes Your On-Call Teammate: Inside Wix's AirBot That Saves 675 Engineering Hours a Month
Wix Engineering15 Jan 2026Company publication

Source summaries

[1]Wix Engineering15 Jan 2026

Wix's engineering blog describes how AirBot investigates Airflow failures and proposes fixes. It gives usage figures for the last 30 days: 4,200 successful flows, 66% positive feedback, 28 of 180 candidate pull requests merged unchanged. It derives 675 saved hours a month from 2,700 interventions at 0.25 hours each.

Full details are in the original sources.

Added 1 Nov 2026 · Updated 1 Nov 2026