What changes when agents arrive

Many lakehouses were designed around people, and one pilot can hide the mismatch. The good news: the architecture decisions that close the gap are ones you can make now.

  • Machine-paced, bursty load: Agents query in bursts nobody scheduled, unlike the dashboards and nightly loads the design assumed.

  • Continuous streams: Fresh data arrives continuously, so ingestion becomes part of the load question.

  • A first strain point you can name: You'll leave able to say where your architecture is likely to strain first as use cases rise.

What it takes to use any engine

“Use any engine” is easy to promise and harder to sustain. The session explains the mechanism, then compares the realistic approaches by trade-off.

  • Learn the mechanism: See how open specification compliance, table maintenance, and streaming ingestion through change data capture work in practice.

  • Weigh four Iceberg approaches: Compare legacy warehouse storage, warehouse-managed, self-managed, and pipeline-managed by trade-off, failure mode, and engineering effort.

  • See one reproducible example: In a published benchmark, ingestion cost $17.46 per day on an open lakehouse pipeline vs. $75.36 per day on an extra-small cloud data warehouse configuration, with 1–3 minute vs. 6–8 minute data freshness. It's an illustration, not a general result.

Source: Qlik, “Benchmarking Ingestion Costs and Performance of Qlik Open Lakehouse Vs a Data Warehouse,” Qlik blog, November 17, 2025. First-party source; “extra-small” framing pending PMM confirmation.

Who should attend

Data architects, plus the heads of data and data engineering leaders who work alongside them. It's a working session, not a product demo or a business case.

Register for the Lakehouse Deep Dive

Join this 30-minute session to explore Apache Iceberg approaches, lakehouse cost trade-offs, and architecture decisions for agentic AI readiness.

SPEAKERS

Ben Saunders
Ben Saunders

Director Product Marketing, Qlik

Simon Swan
Simon Swan

Director Product Marketing, Qlik

Processing...