Kai Wähner

Brokers are disappearing in #ApacheKafka: With #DisklessKafka, all event data lives directly in object storage like S3 or Azure Blob. Lower TCO, elastic scaling, zero data loss DR. The #Kafka protocol remains the unifying layer. #DataStreaming kai-waehner.de/blog/2025/08/1…

Data lineage for real-time streaming is harder than batch. Schemas evolve, pipelines run 24/7, data never stops. Purpose-built solutions + open standards like #OpenLineage are both needed. #ApacheKafka #ApacheFlink #DataGovernance kai-waehner.de/blog/2024/05/1…

#dbt now extends to #ApacheFlink. Data engineers use the same #SQL workflow for batch on #Snowflake or #BigQuery and streaming on Flink. One tool, one team, one governance standard. kai-waehner.de/blog/2026/03/2…… #DataEngineering #DataStreaming

The #AI journey is not just about LLMs. Data quality, #DataGovernance, and exception handling are the real foundation. Great panel at the #EY National AI Conference 2026 in Zurich with 300+ leaders on exactly this reality check. #AgenticAI #EnterpriseAI #Switzerland @EYnews

Shift Left Architecture 2.0: Stream (Kafka), Table (Iceberg), Context (MCP). One event-driven architecture. Three interfaces. A real-time context engine for #AgenticAI works from facts, not assumptions. kai-waehner.de/blog/2026/03/2… #ApacheKafka #MCP #DataStreaming #ShiftLeft

Paid 1,200 USD for a #UFC #VIP ticket in London. The sport was world-class. The premium experience was not. Fan review, business case, and tech vision: kai-waehner.de/blog/2026/03/2… #SportsTech #FanExperience #UltimateFighting #FightNight

Agents hallucinate. Decisions fail. Workflows run on yesterday's data. The problem is not the technology. It is the architecture. #ProcessIntelligence + #EventDrivenArchitecture + #AgenticAI = the Trinity that fixes it.

Stream. Table. Context. The #ShiftLeft Architecture 2.0 adds MCP as the AI interface, giving agents live operational data via a real-time context engine. Kafka + Iceberg + MCP. #ApacheKafka #MCP #AgenticAI #AI

Big News for #ApacheKafka: Schema Registry metadata moves to Kafka headers. Govern your data without touching payloads or breaking existing consumers. From dumb pipes to a smart data plane. #DataGovernance #DataQuality

Kafka moves data. Flink processes it. They complement each other. Not compete. Explained in under 10 minutes on a lightboard: youtube.com/watch?v=PVoc5t… #ApacheFlink #ApacheKafka #DataStreaming

Is a dashboard always the right answer for #ApacheKafka data? Often not. Automation beats visualization. New blog on operational vs. explorative queries and the role of a context engine for dashboards and #AgenticAI: kai-waehner.de/blog/2026/03/1…

In Paris today for "Worth It in Real Time: #EventDrivenArchitecture at l'Oréal." #DataStreaming, operational #AI, and real customer stories. Stale data causes hallucinations. Fresh context from operational systems is what makes AI reliable. #ApacheKafka #Confluent #Paris

Model drift is a silent killer of #AI accuracy. #TikTok beat it with real-time #MachineLearning using #ApacheKafka and #ApacheFlink. Continuous training. No stale data. No batch delays. A great model on yesterday's data is already losing. Read more: kai-waehner.de/blog/2025/02/2…

#ApacheKafka disaster recovery: Asynchronous replication cannot guarantee zero data loss. #SynchronousReplication with Confluent MRC or #WarpStream delivers #RPO0 for mission-critical workloads. Details: kai-waehner.de/blog/2025/08/0…

Big news: KIP-1150 for #DisklessKafka topics accepted by the #ApacheKafka #OpenSource community. Object storage replaces broker disks. Lower #cloud costs, more architecture flexibility. @warpstream_labs leads today. Upstream #Kafka is catching up.

From Texas customer visits this week to Basel on March 25. Joining #Siemens #Healthineers and #Infosys for a #PharmaAndLifeScience roundtable on legacy modernization and real-time #SupplyChain with #DataStreaming. Register free: events.confluent.io/pharmalifescie…

Every major theme at #MWC2026 points to one requirement: real time data flow. #AgenticAI, #5G monetization, network #APIs, autonomous ops. None of it works on stale batch data. #ApacheKafka + #ApacheFlink turn #Telecom ambitions into production reality. kai-waehner.de/blog/2026/03/0…

More and more projects combine #MQTT and #ApacheKafka in #IoT. MQTT handles lightweight device communication. Kafka provides scalable #DataStreaming across the enterprise. Together: end to end from shop floor to IT backend. kai-waehner.de/blog/2023/12/0…

#DataStreaming is replacing #ReverseETL in modern architectures. Batch heavy #DataIntegration drives cost and inconsistency. #ShiftLeft with #ApacheKafka and #ApacheFlink builds trusted #DataProducts in motion, enabling scalable #DataMesh and better foundations for #AgenticAI.

#Tesla is not just a car OEM. It is a software driven #Energy company building a leading Virtual Power Plant. Thousands of batteries and solar panels act as one system via #ApacheKafka. Real time #Telemetry in. Millisecond commands out. More details: kai-waehner.de/blog/2025/02/1…