Articles filed in category 'AI'

  • An AI Stock Analyst That Doesn’t Lie (Probably)

    By: Sahil Malik

    Last updated: Friday, December 26, 2025

    Published in: CODE Magazine: 2026 - Jan/Feb

    Sahil Malik presents a practical blueprint for building an AI-powered stock analyst that aims to deliver up-to-date, verifiable insights rather than false or outdated claims. The article walks through a client-side application that queries Google Gemini with grounding enabled, returns analyzed stock data, and attaches precise, clickable citations extracted from grounding metadata. Sahil emphasizes trust, robust error handling (including exponential backoff), clear user interface design, and citation formatting to enable users to verify every factual claim, arguing that such verifiable AI tools are essential in finance and beyond.

  • Four AIs, One Epic Barbarian Battle

    By: Jason Murphy

    Last updated: Friday, December 26, 2025

    Published in: CODE Magazine: 2026 - Jan/Feb

    Jason surveys four lead AI video models—OpenAI Sora 2, Google DeepMind Veo 3, Kling 2.5 from Kuaishou, and Alibaba Wan 2.5—through their architectural philosophies, tradeoffs, and real-world performance on a cinematic prompt. He argues that beyond specs, each tool reflects a stance on physics, control, motion learning, and multimodal integration, and demonstrates how creators will blend strengths from multiple models.

  • From Commands to Conversations: How AI-Assisted Tooling Is Transforming Angular Development

    By: Sonu Kapoor

    Last updated: Friday, December 26, 2025

    Published in: CODE Magazine: 2026 - Jan/Feb

    Sonu argues that Angular’s Model Context Protocol (MCP) and the Angular MCP Server elevate tooling from mere command execution to intent-aware collaboration. By exposing structured, workspace-aware capabilities via mcp.json, MCP enables AI assistants to reason about a project’s structure, conventions, and best practices, enabling context-driven code generation that aligns with modern Angular patterns (standalone components, Signals, typed forms). Kapoor envisions a future where developers interact with their workspace through intelligent assistants or GUI interfaces, improving reliability, safety, and onboarding while preserving control and auditability.

  • Understanding AI Agents and Agentic AI: Concepts, Tools, and Implementation with SmolAgents

    By: Wei-Meng Lee

    Last updated: Friday, December 26, 2025

    Published in: CODE Magazine: 2026 - Jan/Feb

    Wei-Meng Lee surveys the rise of agentic AI, shifting from passive prompt models to autonomous thinkers that reason, plan, and act using external tools. He introduces SmolAgents as a lightweight framework that lets LLMs orchestrate multi-step workflows, with two core types: CodeAgent for sandboxed Python code execution and ToolCallingAgent for API calls, web searches, and custom functions. Through practical examples and built-in vs. custom tools, Lee demonstrates how agents decompose complex tasks, combine data from various sources, and deliver cohesive, real-time insights, highlighting design choices for real-world applications.

  • What You Need to Know About Fabric

    By: Mike Yeager

    Last updated: Friday, December 26, 2025

    Published in: CODE Magazine: 2026 - Jan/Feb

    Mike argues that Microsoft Fabric is a unifying platform designed to tame fragmented data estates by consolidating diverse data into a single, secure, and governed system called OneLake. By standardizing storage in Delta Parquet across warehouses, lakes, and lakehouses, and pairing robust data engineering, governance (Purview), and AI-enabled Copilots with a simple pay-one-price model, Fabric enables real-time insights, scalable analytics, and AI readiness while maintaining lineage, security, and manageability across dispersed sources and subsidiaries.

  • Using DuckDB for Data Analytics

    By: Wei-Meng Lee

    Last updated: Friday, December 26, 2025

    Published in: CODE Magazine: 2023 - May/Jun

    In this article by Wei-Meng Lee, the author introduces DuckDB, a Relational Database Management System (RDBMS) that supports Structured Query Language (SQL) and is designed for data analytics. Unlike traditional database systems, DuckDB does not require installation and can run queries directly on Pandas data. The article provides examples and demonstrations of how to use DuckDB for data analytics tasks, including loading datasets, querying data using SQL, and performing analytics on the data. The author also discusses the recently added support for JSON ingestion in DuckDB. Overall, the article highlights the convenience and efficiency of using DuckDB for data analytics tasks.