DATA Pill #196 – MCP vs Skills, Lakehouse Architectures, Text-to-SQL Agents & the AI Model Avalanche

ARTICLES

Two emerging approaches for extending LLM capabilities are MCP tools and agent skills. MCP provides a standardized interface to external systems such as APIs, files, or databases, enabling structured and reliable operations. Skills, on the other hand, are lightweight behavioral instructions that guide how agents reason and act. The comparison highlights the trade-offs between structured integrations and flexible prompt-based capabilities when building agentic systems.

Building production ML systems requires coordinating feature management, experimentation, and model deployment. This architecture combines Feast for feature storage and retrieval, MLflow for experiment tracking and model registry, and Kubeflow for orchestration. Together they create a modular open-source stack that helps teams move from feature engineering to production pipelines with consistent workflows.

The lakehouse architecture combines the flexibility of data lakes with the reliability of warehouses. By using open table formats on object storage and separating compute from storage, organizations can support BI, streaming, and machine learning on the same platform. Modern lakehouses rely on open catalogs, multi-engine compute, and governance layers to avoid vendor lock-in and enable scalable data platforms.

Text-to-SQL systems often struggle with understanding user intent and navigating large data catalogs. A unified embedding approach captures both business intent and dataset context, allowing queries to be mapped to relevant tables and historical analyses. The system also leverages governance signals and previous queries to improve accuracy and reliability for analytical workflows.

The first week of March 2026 brought a wave of AI model releases across multiple domains, including language models, video generation systems, and multimodal tools. Updates include larger context windows for LLMs, more efficient small models capable of running on edge devices, and new approaches to video generation and editing. The trend highlights rapid progress across both frontier models and open-source alternatives.

NEWS

Meta has acquired Moltbook, a social platform where AI agents interact and share content in topic-based communities. The platform gained attention for hosting millions of autonomous bots communicating with each other. The acquisition signals growing interest in ecosystems where agents interact, discover tools, and collaborate across networks.

Amazon Connect Health introduces AI agents designed to automate administrative workflows in healthcare. The system can verify patients, schedule appointments, retrieve records, generate clinical documentation, and summarize visits. By integrating directly with electronic health record systems, the platform aims to reduce operational overhead while improving patient experience.

DATATube

We present LingBot-World, an open-sourced world simulator stemming from video generation. Positioned as a top-tier world model, LingBot-World offers the following features. (1) It maintains high fidelity and robust dynamics in a broad spectrum of environments, including realism, scientific contexts, cartoon styles, and beyond. (2) It enables a minute-level horizon while preserving contextual consistency over time, which is also known as "long-term memory". (3) It supports real-time interactivity, achieving a latency of under 1 second when producing 16 frames per second.

AI coding tools often generate plausible code that passes automated checks but introduces long-term maintenance issues. Learn to identify five specific patterns of this hidden technical debt. A structured three-question review framework helps teams maintain quality and velocity.

Explore how AI is shifting industry hiring patterns and compressing the distance between idea and execution. Discover why traditional software roles are evolving towards oversight and architectural design.

TOOLS

MyClaw provides managed hosting for OpenClaw, a free open‑source AI assistant. Without MyClaw, users must clone repositories, manage dependencies and resolve environment conflicts; with MyClaw, a dedicated private OpenClaw instance runs 24/7 without setup

CONFS, EVENTS, WEBINARS & MEETUPS

This live webinar will discuss infrastructure patterns for low‑latency, real‑time AI and ML workloads. Experts from Dragonfly and The New Stack will share strategies for predictable performance, efficient memory management and avoiding runaway costs when serving models at scale.

this meetup celebrates the release of Data Pipelines with Apache Airflow (Second Edition). Attendees will hear about Airflow 3’s new features, learn about the book‑writing process, participate in a quiz and enjoy networking with the authors

A half‑day event covering blueprints for building an open data lakehouse on AWS and applying agentic AI to data engineering. Sessions will explore reference architectures (Iceberg + Glue + multi‑engine compute), governance patterns, integrating AI into dbt and orchestration workflows, and designing conversational data agents. Seats are limited

PINNACLE PICKS