World Congress 2025 Aug 20, 2025 Session details

Bringing Clarity to Event Streams: Enabling Analytics and AI Through Rich Metadata

Clemens Vasters

Feeding raw event data to LLMs leads to hallucinations. A schema-first approach provides the exact machine-readable context needed to generate accurate, predictable code for your event pipelines.

Pause
Mute Enter Fullscreen
#1 about 5 min

Challenges of interpreting raw data with language models

Without physical context or schema definitions, language models make unpredictable assumptions about raw values like temperature units.

#2 about 5 min

Enhancing prompt output through explicit schema constraints

Providing framing prompts with precise standard schema definitions ensures accurate interpretation and reliable generation of data records.

#3 about 2 min

Using independent schemas to drive polyglot code generation

Treating schemas as independent primary assets rich with human-readable context significantly improves polyglot code generation workflows.

#4 about 2 min

Evaluating inconsistencies in model-generated event stream formats

Relying on raw prompts to construct event objects results in chaotic, non-standardized formats across distributed application components.

#5 about 2 min

Standardizing generated messages via CloudEvents metadata parameters

Supplying formal specifications as instruction constraints configures large language models to produce natively compliant and uniform event structures.

#6 about 5 min

Generating reliable publisher clients using composite endpoint metadata

Combining foundational event declarations with broker configurations guarantees the instant output of accurate publisher clients across multiple languages.

#7 about 2 min

Replacing tribal knowledge in complex event streaming pipelines

Formalizing metadata definitions eliminates tribal assumptions by explicitly documenting structural contracts between isolated stream producers and consumers.

#8 about 7 min

Enforcing streaming data contracts within enterprise analytics infrastructure

Capturing event stream definitions into strict physical repositories allows enterprise platforms to aggressively drop non-compliant infrastructure data.

#9 about 3 min

Adopting JSON Structure as a strict data definition language

The new JSON Structure specification delivers primitive data boundaries and robust namespace rules to transcend validation-focused methods.

#10 about 4 min

Establishing universal metadata ecosystems with xRegistry and CloudEvents

Coupling standardized package registries for discovering endpoints builds an interoperable metadata graph vital for deterministic application integration.

Matching moments

3:17 min

Establishing metadata standards for real-time data pipelines

Clemens Vasters Clemens Vasters · WWC Europe 2026

1:39 min

Standardizing messaging metadata for artificial intelligence and cloud

Clemens Vasters Clemens Vasters · WWC 2025

4:30 min

Introducing data management and the shift to streaming

Mary Grygleski Mary Grygleski · LIVE

2:12 min

Applying JSON metadata standards for digital asset interoperability

Michiel Mulders Michiel Mulders · WWC 2024

4:43 min

Defining complex metadata schemas for microfrontend APIs

Alexander Günsche Alexander Günsche · LIVE

4:23 min

Clarifying terminology around different event computing software forms

Mary Grygleski Mary Grygleski · LIVE

Upcoming sessions on this topic

Open session

World Congress 2026 North America

From Guesswork to Governance: Data Contracts Bring API Discipline to Apache Kafka

Sandon Jacobs

Senior Developer Advocate at IBM

Sandon Jacobs
Open session

World Congress 2026 North America

Beyond SQL Generation: How to Teach Agents What Your Database Actually Means

Kris Jenkins

Lead Developer Advocate

Kris Jenkins
Open session

World Congress 2026 North America

Bluesky's Open Source Moderation Tools: LLM-based Event Detection in Python

Alex Garnett

Developer Relations, Bluesky

Alex Garnett
Open session

World Congress 2026 North America

You Can’t Re-Run Sunlight: Designing ML Data Architectures for Physical AI

An Phan

Senior Data Infrastructure Engineer @ Hippo Harvest

An Phan
Open session

World Congress 2026 North America

AI Agents are Only as Smart as their Context: Building a Real-Time Context Engine at Intuit

Bharat Patel

Lead Software Engineer at Intuit

Bharat Patel
Open session

World Congress 2026 North America

No Single Model to Rule Them All: Building Resilient AI Agents Across Open & Closed LLMs

Emmanuel Acheampong

Senior Manager Developer Relations at Crusoe AI

Emmanuel Acheampong