- Big Data News Weekly
- Posts
- 🤖 Why AI Coding Agents Make Bad Analytical Decisions
🤖 Why AI Coding Agents Make Bad Analytical Decisions
🦾Plus: 🏦 Perplexity plugs its AI agent into bank accounts
In partnership with
Hey folks! Let’s get into Big Data and AI craziness…
In today's edition: Oxford's AI catches heart failure five years early. Amazon's $200B AI spend has receipts. And the data engineer's roadmap gets its 2026 update. 👇
👨💻The Data Engineer’s Roadmap
📐Six (and a half) intuitions for KL divergence
🗂️What Category Theory Teaches Us About DataFrames
🚀 Making Claude Cowork ready for enterprise
💻 OpenAI Takes Aim at Claude with a $100 Coding Tier
💡 AI Tutorial:How to do live video calls with AI agents
🤖AI Tools and Data Tools to checkout
Claude Cowork started as a way to bring the agentic power of Claude Code to non-developers. Now Anthropic is making it viable at the enterprise level. The update introduces role-based access controls for Enterprise plans so admins can assign specific Claude capabilities to different teams via SCIM or manual setup. Group spend limits let companies set per-team budgets from the admin console.
AI agents now read your docs almost as much as humans do.
Mintlify analyzed 790 million requests across its documentation platform. The finding: AI coding agents account for 45.3% of all traffic, nearly tied with traditional browsers at 45.8%.
Two tools are driving almost all of it:
Claude Code: 25.2% of total traffic, more requests than Chrome on Windows
Cursor: 18% of total traffic
Together they account for 95.6% of all identified AI agent traffic
The rest of the field, OpenCode, Trae, ChatGPT, and NotebookLM, is showing up but nowhere close.
One caveat: OpenAI's Codex doesn't send an identifiable user-agent header, so the real agent percentage is likely even higher.
The takeaway for anyone maintaining developer docs: your documentation now serves two audiences. Structure and machine-readability matter as much as clarity for human readers.
Data engineering is fascinating, if not fulfilling career. You are at the helm of every business operation that requires data, and as long as users generate data, businesses will always need data engineers. In other words, job security is guaranteed.
KL-divergence is a topic which crops up in a ton of different places in information theory and machine learning, so it’s important to understand well. Unfortunately, it has some properties which seem confusing at a first pass (e.g. it isn’t symmetric like we would expect from most distance measures, and it can be unbounded as we take the limit of probabilities going to zero).
Every dataframe library ships with hundreds of operations. pandas alone has over 200 methods on a DataFrame…Without a framework for telling them apart, you end up memorizing APIs instead of understanding structure…I ran into this question while building my own dataframe library. I needed to decide which operations were truly fundamental and which were just surface-level variations.
AI coding agents write correct code. They make unreliable analytical decisions. That gap between syntactically valid code and statistically sound conclusions is where costly mistakes are made. Decision Lab is an open-source framework that closes this gap by encoding domain expertise, multi-path exploration, and Bayesian uncertainty quantification into every agent run…
Will Your Retirement Income Last?
A clear retirement income plan starts with knowing your costs and building a portfolio that can meet them. Fisher Investments' Definitive Guide to Retirement Income helps investors with $1,000,000 or more structure a strategy built to last.
👨💻 Data Tools, Libraries
Homebrew 5.0.0 is now available. The most significant changes since 4.6.0 are download concurrency by default, official support for Linux ARM64/AArch64, timescales for deprecating macOS Intel, and the removal of macOS Gatekeeper bypass behaviours.
Stripe AI (GitHub Repo)
This repository contains a collection of software development kits for integrating Stripe with large language models and agent frameworks.
AI News:
Perplexity just rolled out a new Plaid integration that lets users connect bank accounts, credit cards, and loans directly to its Computer agent, turning it into a full personal finance hub. Plaid's 12K+ bank network feeds into Computer, with users able to pull in checking, credit, loan, and brokerage data for a read-only view of their money.
200+ AI Side Hustles to Start Right Now
AI isn't just changing business—it's creating entirely new income opportunities. The Hustle's guide features 200+ ways to make money with AI, from beginner-friendly gigs to advanced ventures. Each comes with realistic income projections and resource requirements. Join 1.5M professionals getting daily insights on emerging tech and business opportunities.
The AI chatbot now answers questions with interactive visualizations directly in chat. To try it out, just ask Gemini Pro to “show me” or “help me visualize” a topic. Gemini also got updated with notebooks: dedicated workspaces where you can group chats, upload files, or add instructions. It’s rolling out now, starting with paid accounts.
Amazon CEO Andy Jassy shared his annual shareholder letter with the company's first-ever AI revenue figures and a defense of the $200B planned capex, dismissing bubble talk and floating the idea of selling Trainium chips to outside buyers. Amazon's $200B AI spending rattled investors this year, with Jassy's letter firing back with first-ever revenue figures and locked-in customer demand.
OpenAI now offers a $100/month plan sitting between its $20 Plus tier and $200 Pro tier. The new plan is designed for developers who rely heavily on Codex but found Plus too limited and the $200 plan too expensive. The $100 tier offers 5x the Codex capacity of Plus. The $200 plan still exists and provides 20x Plus limits for the most demanding workflows.
Researchers at the University of Oxford introduced an AI system that picks up invisible changes in heart fat from routine CT scans, flagging patients at high risk of heart failure up to five years out — with 86% accuracy across 72K patients. Fat around the heart shifts texture when the muscle beneath is inflamed, with the AI reading the patterns invisible to doctors on any current scan.
AI Tutorial
How to do live video calls with AI agents
Image Source: Pika
Go to Pika AI Self and create your living avatar.
For other agents, you can download the Skill on GitHub here.
Go to ‘Skills and Connections’ > ‘Real-time Video Chat’ > ‘Try Now’
Now simply invite them to a Google Meet using a meeting link.
Your Pika Self will join the call and carry on conversations with your personality and taste.
AI can help you move faster, but real leadership still requires human judgment.
The free resource 5 Traits AI Can’t Replace explains the traits leaders must protect in an AI-driven world and why BELAY Executive Assistants are built to support them.
🔥Top AI tools to increase productivity:
Zoice is the single platform for every creator. Transcribe, generate, and animate your content.
Floowed is a flexible, no-code AI credit workflow automation platform
BookSwift is a modern appointment booking platform for providers
Marketsy.ai provides a smart e-commerce experience supported by a powerful admin panel.
StrideFuel - Built for weight loss success—especially GLP-1 users
WorldEngen is an AI copilot for 3D production that helps professional teams
AppWizzy is an AI tool that helps you build and host full-stack web applications
SongGuru.AI: An AI-Based Music Creation and Audio Processing Platform
View our database of all the best AI tools for your needs: aitoolsup.com
Have cool resources to share? Submit AI tool
A.I. Generated Image of the Day
👀 Age of Emperors
Recommended reading:
SPONSOR US
Get your product in front of Big Data & AI enthusiasts
Our newsletter is read by thousands of tech professionals, investors, engineers, managers, and business owners around the world.
Interested in Sponsoring the Big Data News Weekly Newsletter?Get in touch today
What did you think of today's email?Your feedback helps me create better emails for you! |