Data &amp; AI Infrastructure

What Is Context Engineering? A Field Guide to the AI Context Window

Redis's Jessica Emmons explains context engineering, the four most common context failures, why RAG falls short, and how context engines work.

Agentic Data Engineering: How Matillion's Maia Builds Data Pipelines

At COLLIDE 2026, Matillion's Arawan Gajajiva showed how the Maia AI agent plans, builds and tests data pipelines with a human in the loop.

From Raw Data to Predictions: How BigQuery AI Speeds Up ML Workflows

Google Cloud's Shan Kulandaivel shows how SQL-based forecasting, prediction and document AI in BigQuery move ML ideas to production faster.

Why AI Fails Without Data Products and Data Contracts

Actian CTO Emma McGrattan explains how data products and data contracts help enterprises move AI from pilot to production.

Data and AI Control Plane: Why AI Agents Need One Identity and Boundaries

NexusOne CEO Anu Jain on why AI agents need one identity, complete context and hard boundaries, and why enterprises should own their ontology.

Streamhouse Architecture: Real-Time Data for Smarter AI Agents

IBM's Ann Leach explains why stale data limits AI agents, what a streamhouse is, how it complements the lakehouse, and how Ferrari used real-time data.

Single Source of Truth for Agentic AI: From Databases to Meaning

SAP, Turgon, Skyflow and AtScale leaders on how agentic AI redefines the single source of truth: semantics, runtime controls and evals on your own data.

Real-Time Data for AI Decisions: Lessons From IBM's Client Zero

IBM's Tim Richer on why stale data stalls AI, the three use cases scaling fastest, and lessons from IBM's internal Ask IBM agents.

The Self-Funding AI Strategy: How Verisk Lets AI Pay for AI

Revefi's Girish Bhat and Verisk CDO Louis Dimodugno explain how cutting data and AI waste can fund new AI work, plus a 4-step playbook.

AI-Ready Data: Why Enterprises Are Bringing AI to Their Data

Everpure's Kaycee Lai explains what AI-ready data means, why unstructured data blocks production AI, and how to run AI where data lives.

Sovereign AI Architecture: 7 Layers for Production-Ready AI Agents

Fulcrum Digital founder Rajesh Sinha explains a seven-layer sovereign AI architecture and why inference routing makes or breaks production AI.

AI Agents Need Context, Not Just Data: Why Data Lakes Fall Short

Teradata's Ben Sanders explains why data lakes built for humans fail AI agents, and the six kinds of context agents need to deliver ROI.

Agentic AI Demos: Snowflake Cortex Code and Precisely's MCP Server

Snowflake demos spec-driven development with Cortex Code, and Precisely shows an AI assistant pulling location risk data through MCP.

AI Demo Day: What Mistral AI, SAS, Fivetran and Precisely Showed

Four enterprise AI demos: Mistral document and voice AI, SAS forgery detection, Fivetran's one-copy data lake and Precisely location data.

Agentic-Ready Data: Why AI Agents Need Metadata, Not Just Clean Data

Snowflake and Precisely experts explain why data built for dashboards fails AI agents, and how metadata, semantic layers and lineage make it agent-ready.

Scaling AI Architecture: How to Build AI Systems That Grow Reliably

SAP, Skyflow and Fivetran leaders explain the isolation, modularity, evaluation and governance patterns that take AI from proof of concept to scale.

AI Workflows in Claude: Supermetrics and Precisely Demo Read-Write AI

Supermetrics and Precisely demo Claude workflows that analyze ad performance, launch campaigns and fix SAP master data through an MCP server.

AI Agent Observability: Six Early Warning Signals for Production Agents

Monte Carlo and Snowflake on AI agent observability: why agents degrade after launch, a four-layer trust framework and six early warning signals to track.

The Context Layer for Enterprise AI: Metadata, Semantics and Memory

Collate, Graphwise, Precisely and AtScale debate the AI context layer: metadata graphs, semantic layers, knowledge graphs, RAG, memory and orchestration.

Why AI Pilots Fail to Scale, and How Enterprises Get to ROI

Snowflake, Revefi and Coca-Cola AI leaders explain why AI pilots stall, how to choose what to scale, and the 70-20-10 rule.

How to Build AI Agents With MCP Tools: Teradata and Coder Live Demos

Teradata builds an agent that reasons over policies and tables via governed MCP tools; Coder shares agent harness, sub-agent and context lessons.

AI Infrastructure at Scale: Semantic Layers, GPUs and Inference Costs

AtScale and Rhino Federated Computing leaders explain the infrastructure layers, governance and inference cost controls production AI and agents need.

Agent-Ready Data Architecture: Streaming Real-Time Data to AI Agents

Striim and Google Cloud explain why batch warehouses fall short for AI agents and how CDC, event streaming and MCP deliver fresh context on Google Cloud.

Four Signs Your Data Is Ready for AI: Real-Time, Governed, Event-Driven

Striim and Google Cloud explain four signs of AI-ready data: current state, derived not accessed, preserved business meaning and event-driven design.

AI Demo Day Q1: Mistral, Google Cloud, Aerospike, Striim and More

Mistral AI, Google Cloud, Aerospike, Precisely, Striim and PromptQL showed agents, churn analysis, real-time nudges, location context and data masking.

AI Prototype to Production: What Makes Enterprise AI Apps Production-Ready

DataHub, Precisely, Snowflake and Fivetran leaders explain why AI prototypes stall, what enterprise-grade means, and why governance must start on day one.

Agentic Data Management: How AI Agents Let Analysts Self-Serve Data

Matillion's team shows how the Maia agent lets analysts and data scientists build governed pipelines and fix failures without waiting on data teams.

AI-Ready Data: Why AI Fails on Foundations, Not Algorithms

Fivetran, Snowflake and Accenture leaders explain what AI-ready data means and how quality, access and governance decide whether AI projects succeed.

Real-Time AI Fraud Detection: Why Milliseconds Matter

Aerospike's Sesh Seshadri explains how ML, graph analytics and multi-stage inference stop financial fraud within a 50 ms window.

Measuring AI ROI: How to Tie Every Use Case to a Business OKR

Leaders from The Home Depot, Mars Wrigley and Equifax explain how to measure AI ROI, balance speed and cost, and set executive expectations.

Custom MCP Servers and the Composable, Agentic Enterprise Platform

SnapLogic's Dylan Vetter explains how integration platforms prepare data for AI, democratize integration with governance and build custom MCP servers.

The AI Production Gap: Why Bad Data, Not Models, Blocks AI at Scale

Syncari's Jack Smith explains why integration sprawl, data lakes and legacy MDM stall AI pilots, and how a centralized agentic MDM layer helps them scale.

Vibe Analytics for EDA: Turning Weeks of Chart Requests Into Minutes

Broadband Insights CEO Patrick Hutto explains how vibe analytics with Plotly AI Studio speeds exploratory data analysis, mockups and self-service.

Agent Mesh: How to Build an Enterprise Platform for AI Agent Collaboration

Grid Dynamics' Stas Stolpovskiy explains the agent mesh: what to control centrally, top-down vs bottom-up platforms, and why A2A beats MCP there.

Five Questions Enterprises Must Answer to Scale Agentic AI, per IBM

IBM's Ryan Zombo explains how agents, trusted data and centralized governance take enterprise AI to production, and shares IBM client zero results.

High-Performance Data Infrastructure for AI: Aerospike's Cost Case

Aerospike CTO Srini Srinivasan explains how hybrid memory cuts AI infrastructure costs and why agentic AI multiplies real-time database load.

Data Science Agents in BigQuery: Agentic AI for Data Scientists

Google Cloud's Suda Srinivasan demos a data science agent building a bike-share demand forecast in BigQuery and explains what agents can't replace.

Aerospike Live Demo: Surviving Node Failure and Catching Fraud by Graph

At COLLIDE 2025, Aerospike's Stan Christ shut down cluster nodes under load, then showed graph-based fraud detection flagging transfers many hops away.

Agentic AI Showcase: Supermetrics, Striim and Lumi AI Demo Data for Agents

Supermetrics, Striim and Lumi AI demo how AI agents get trusted data: marketing data via MCP, real-time CDC with masking, and an AI data analyst.

Choosing AI Use Cases That Pay Off: A Data Executive's Guide to ROI

Leaders from AtScale, Snowflake, Denodo, Fivetran and Revefi on telling AI hype from value, build vs. buy, board pressure and use cases with proven ROI.

GenAI Demo Day Q3: Snowflake, Denodo, Zenlytic, PromptQL and More

Snowflake, Denodo, Zenlytic, PromptQL, Fivetran, Google Cloud and Precisely demo AI agents, semantic context, A2A and MCP servers in 10 minutes each.

One Real-Time Data Layer for Predictive, Generative and Agentic AI

Aerospike's Sesh Seshadri explains multi-step inference, why recommendations, search and RAG share one pipeline, and the data agentic AI needs.

Multimodal AI Search: How Tensors Make Video Retrieval Work at Scale

Vespa.ai's Bonnie Chase and Zohar Nissare-Houssen explain why vectors aren't enough for multimodal AI and demo hybrid video search with tensors.

Data Lakehouse Architecture for AI: Governance, Cost and Migration

Fivetran, General Mills and Webster Bank leaders explain how data lakehouses, Apache Iceberg and open catalogs support governed, cost-effective AI.

GenAI Demo Day Q2: Snowflake, Google, Precisely, Zenlytic and More

Seven GenAI demos: Snowflake Cortex Agents, Gemini 2.5 on Vertex AI, Precisely Data Graph, Zenlytic, Fivetran, Dataiku and Zerve's data science agent.

What Is an AI Factory? NVIDIA and HPE on Scaling Enterprise AI

NVIDIA and HPE leaders explain AI factories: compute, networking and software as a relay team, private cloud AI options, liquid cooling and AI strategy.

How AI Augments HPC Simulation: Physics Meets Probability

HPE's Steve Heibein explains how AI narrows the search space for physics-based HPC simulation, from COVID research to NASA parts, and why cooling matters.

Sovereign AI for Government: Data, Operations and Air-Gapped AI

NVIDIA and HPE leaders define sovereign AI, the three pillars of digital sovereignty and the challenges of deploying AI in air-gapped environments.

Enterprise GenAI Applications: Security, Accuracy and Evaluation at Scale

AtScale, Nexla, Informatica, Vespa and Skyflow leaders explain how to secure data, reduce hallucinations and evaluate enterprise GenAI apps at scale.

Data + AI Observability: The Four Root Causes That Block AI Readiness

Monte Carlo's Shane Murray and Michael Segner explain how data, system, code and model failures break AI and how AI agents help monitor data quality.

GenAI Demo Day: Snowflake Cortex Agents, TiDB GraphRAG and Zenlytic

Snowflake, TiDB and Zenlytic demo AI agents over enterprise data: Cortex Agents, GraphRAG in a single database and open-ended AI analytics.

Production Python Data Apps: How Broadband Insights Scaled With Dash

Broadband Insights CEO Patrick Hutto shows how Python data apps on Plotly Dash Enterprise helped small ISPs plan networks and cut install times.

The Hidden Costs of AI: Scaling ML From Notebook to Production

Tecton, Mosaic Health and JPMorgan Chase leaders explain the people, pipeline and premature-optimization costs of scaling ML to production.

Real-Time AI Architecture: Fresh Features, Low Latency, Lower Cost

Tecton, Sam's Club and AT&T panelists explain when AI needs real-time data, how to build fresh low-latency features, monitor drift and control costs.

ColPali for Multimodal RAG: Retrieving Charts, Tables and PDFs Visually

Vespa.ai CEO Jon Bratseth and RavenPack's Aakarsh Ramchandani explain how ColPali retrieves PDF pages visually, its limits and how to scale it.

Scalable RAG Systems: Cutting Cost and Securing Enterprise Retrieval

Tecton, Nexla and Novartis experts on scaling RAG: batch vs. real-time serving, smaller models, index sync, chunk-level access control and evaluation.

Large Quantitative Models: How Physics-Based AI Speeds Materials R&D

SandboxAQ's Ang Xiao explains how large quantitative models pair AI with physics simulations to speed battery, chemistry and materials discovery.

From Prototype to Production AI: Cutting Tech Debt and Train-Serve Skew

Tecton, Emory Healthcare and BBDO leaders on moving AI to production: tech debt, standardization, train-serve skew, automation and governance.

Future-Proof AI Infrastructure: Self-Hosted LLMs, Governance and Lock-In

SandboxAQ, Alteryx and Plotly engineers on self-hosted vs. cloud LLMs, governed AI access, data quality testing and avoiding vendor lock-in.

Fraud Detection Build vs. Buy: How North Built Real-Time ML In-House

North's Ben Orkin explains why the payment processor replaced a third-party fraud tool with its own real-time ML system and feature store.

AI ROI Across the Project Life Cycle: Business Cases, Data and Cloud Costs

Bank of America, Carelon Health, Lockton and PRGX leaders on AI business cases, the three E's of ROI, data bottlenecks and controlling cloud costs.

Data Democratization With PII Controls: Piedmont and Home Depot

Piedmont, The Home Depot, Alteryx and BBDO on democratizing data: certified data sets, business-owned access approvals, break-the-glass PHI logging.

AI Strategy Starts With Data Strategy: Southern Company, Clorox, Tecton

Southern Company, Clorox, Morgan Stanley and Tecton on aligning data and AI strategy: data quality, features as code, scaling pilots and real-time data.

GenAI Demo Day: Rafay, AnswerRocket, Crowdbotics and DataRobot Demos

Four generative AI demos: Rafay self-service GPUs, AnswerRocket's Max analytics assistant, Crowdbotics AI requirements and DataRobot apps with guardrails.

Relational Deep Learning: Graph Neural Networks on Snowflake Data

Kumo AI's Hema Raghavan explains relational deep learning: turning Snowflake tables into graphs so neural networks replace manual feature engineering.