In 2013, data analysis at Wayfair was trapped on on-premise SQL Server boxes that couldn't even join across databases. Over the following decade, cloud data warehouses and the modern data stack tore down constraint after constraint, turning data scientists from reporters into operational decision-makers. 📈 Now AI has made producing an analysis nearly free. But in "The Shape and Feel of the Post-AI Data Stack," Ian Macomber argues that agreeing on reality never gets cheaper. The more dashboards proliferate, the more likely different interfaces return different answers to the same question — a growing "consensus divergence" problem. 🤖 The post-AI data stack Macomber describes puts an agent harness between infrastructure and people: agents parse data products and reassemble insights for humans, not the other way around. That demands four things — artifacts agents can read, tools agents can operate via API, context that stays agent-agnostic across vendors, and consensus that can be systematically tested. Ramp, he notes, fires identical questions through every interface and measures how often the answers diverge, aiming for zero. ✍️ Macomber's conclusion: a data scientist's worth will be measured not by any single analysis, but by how well their judgment gets encoded into infrastructure so every future agent, decision, and employee inherits what's true and why. URL: #DataStack# #AIAgents#

Three-person data team, a queue of weekly requests, and every analytical question outside predefined dashboards stuck in someone's inbox — that was LangChain's BI era. 🔍 The team decided to change. What they needed was a platform unifying dashboards, notebooks, and conversational interfaces with native AI agent capabilities. They chose Hex and designed a five-layer context architecture: dbt data model definitions, a semantic layer, workspace guides, endorsements as trust signals, and GitHub integration. The migration was complete in six weeks with 100% company adoption. The transformation's core wasn't technical — it was about explicitness. Rewriting a weak definition like "account_status: The status of the account" into a full description covering lifecycle states, default filters, and reporting conventions alone changed how reliably the agent answered questions. Key metrics — ARR, pipeline, customer health — were given single authoritative definitions in the semantic layer. Endorsements pointed agents to canonical sources wherever multiple assets addressed the same concept, preventing confusion before it could happen. Today, every function — marketing, product, sales, customer engineering — runs its own analysis without routing through the data team. Monthly agent conversations total roughly 2,200, processing 40x the volume the three-person team could handle manually. The team's role has shifted from "answer every question" to "design the system that lets others answer questions." LangChain's detailed account of this journey, "How LangChain Built an Agent-First Data Stack," makes the case — in concrete numbers and design principles — for what a data team looks like in the agent era. #DataStack# #AIAgent#

Your AI agents are only as good as the data behind them. Fragmented data, inconsistent definitions and weak governance can hold marketing teams back. Learn how leading organizations are preparing for the agentic enterprise in our latest report – the 5th edition of the Modern Marketing Data Stack.👇

CoinGecko API is now the most complete AI-native crypto data stack for builders & AI agents. 🤖 Over the past few months, we released 4 AI tools to help you build, ship, & analyze faster: • CLI • CoinGecko Skill • x402 Endpoints • MCP Server Here's how they work 🧵

TechCrunch reported that Encord is working with Zander Labs to test a new physical-AI data pipeline: capturing first-person video and brain-wave signals while humans perform manipulation tasks. This is not about “mind-controlled robots.” The goal is to make training data more informative. Brain signals could help mark intent, error, surprise or cognitive load — context that video alone may miss. Encord’s broader physical-AI data stack also includes in-field collection, teleoperation, multi-sensor data, annotation and deployment feedback. Robots need structured examples of hands, tools, objects, motion, timing and failure cases to learn the physical world.

AI agents are turning $NBIS backed ClickHouse into one of the fastest-growing pieces of the data stack with ARR surpassing $350M. OpenAI usage has reportedly jumped ~10x to more than 30 petabytes per day while shifting parts of its log workload from $DDOG.

U.S. Bank just ran its own stablecoin on a public chain. This leaves a record anyone can read. Next step: get those records into the bank's data stack in a shape an auditor accepts. Amp does that part.

Governance isn't slowing innovation, it’s enabling it. Snowflake CMO Denise Persson unveils the 5th Modern Marketing Data Stack report. Get the blueprint for the agentic enterprise. 👉🏻

Introducing PostgresBench, a benchmark to compare managed Postgres services. Postgres handles transactional workloads, while ClickHouse handles analytical workloads. Together they form a unified data stack enabling a "best-of-breed" foundation SaaS and AI applications.