BetaMaShop is in public beta. We improve it continuously, and your feedback shapes what comes next.
MaShop Journal

The MaShop blog

A look at what the MaShop team is making lately.

⌘ K
This issue · 8 min
IndustryAUG 14BY MASHOP

Dynamic pricing for a small shop, and the line not to cross

Repricing by stock and season is ordinary trade. Pricing by who is looking is not. What the 30 day reference rule does to your discount badge.

M
MaShop
MaShop team
8 min
This week

What we wrote this week.

We write when we have something to say, not when the calendar says.

Archive

Everything we have written so far.

By MaShop team.

007

Machine Translation Is Fluent Now. That Is the Problem

ComparisonsAUG 13MaShop5 min
Comparisons
008

AI Resume Screening Is a Compliance Purchase, Not a Time Saver

IndustryAUG 13MaShop5 min
Industry
009

Your Forecast Has to Beat Last Year, and Usually It Does Not

ResearchAUG 13MaShop5 min
Research
010

AI Act Compliance, Mapped Obligation by Obligation

IndustryAUG 13MaShop5 min
Industry
011

An AI Phone Agent Is Judged on the Pause, Not the Voice

ToolsAUG 13MaShop5 min
Tools
012

Agentic Shopping Is Under 1% of Etsy's Traffic

IndustryAUG 13MaShop5 min
Industry
013

Who Owns an AI Assisted Track? The Line Is Human Authorship

IndustryAUG 13MaShop5 min
Industry
014

The Bot Took Notes and Left Out the Decision

ToolsAUG 12MaShop5 min
Tools
015

An AI Policy Your Staff Will Actually Follow, in One Page

IndustryAUG 12MaShop5 min
Industry
016

The AI Text on Your Website Now Carries an Invisible Mark

IndustryAUG 12MaShop5 min
Industry
017

Ask Excel's AI for a Formula, Not for the Answer

ToolsAUG 12MaShop5 min
Tools
018

Your Catalogue Has Thousands of Images. Which Need Alt Text?

ToolsAUG 12MaShop5 min
Tools
019

Getting Line Items Out of a Thousand Supplier PDFs

ToolsAUG 12MaShop5 min
Tools
020

The Things an AI Sales Agent Will Not Do for You

ComparisonsAUG 12MaShop5 min
Comparisons
021

Legal AI Sorted by Task, and by Who Pays for the Error

IndustryAUG 11MaShop5 min
Industry
022

Four Water Numbers for the Same Data Center

IndustryAUG 11MaShop5 min
Industry
023

An AI Agent Was Told to Book a Class. It Cancelled a Stranger.

IndustryAUG 11MaShop5 min
Industry
024

The Parts of a Spec an Agent Actually Acts On

ToolsAUG 11MaShop5 min
Tools
025

You Approve 97% of AI Prompts. That Is Not Oversight

IndustryAUG 10MaShop5 min
Industry
026

Governance Without an Inventory Is Theatre

IndustryAUG 10MaShop5 min
Industry
027

The Context Window Is a Budget. Here Is How to Spend It

ToolsAUG 10MaShop5 min
Tools
028

What Actually Fits on the Device, and What Cannot

ToolsAUG 10MaShop5 min
Tools
029

AI Moved the Work From Producing to Checking

ToolsAUG 10MaShop5 min
Tools
030

MLOps as Three Levels, and the One Thing to Build Next

ToolsAUG 10MaShop5 min
Tools
031

What Actually Moves Your Citation Rate in AI Answers

IndustryAUG 10MaShop5 min
Industry
032

AI Grading Works for Comments, Not for Scores

ResearchAUG 06MaShop5 min
Research
033

AI in Healthcare, Graded by the Evidence Behind It

ResearchAUG 06MaShop5 min
Research
034

AI Engineer Salary: Why No Two Sources Agree

IndustryAUG 06MaShop5 min
Industry
035

How to Tell a Real Robot Demo From a Staged One

ResearchAUG 06MaShop5 min
Research
036

The EU AI Act as a Decision Tree and a Timeline

IndustryAUG 06MaShop5 min
Industry
037

How a Transformer Reads a Sentence, Step by Step

ResearchAUG 06MaShop5 min
Research
038

A Court Said an Agent Can Shop on Your Site

IndustryAUG 06MaShop5 min
Industry
039

What Is an LLM? From Text to an Answer in Five Stages

ResearchAUG 05MaShop5 min
Research
040

Ads Moved Into the AI Shopping Assistant

IndustryAUG 05MaShop5 min
Industry
041

Chain of Thought Prompting, and When It Stops Helping

ToolsAUG 05MaShop5 min
Tools
042

Conversational AI, Rung by Rung

ComparisonsAUG 05MaShop5 min
Comparisons
043

The AI Bubble, Argued With Numbers

IndustryAUG 05MaShop5 min
Industry
044

Reading a ComfyUI Workflow as a Graph

ToolsAUG 05MaShop5 min
Tools
045

AI Tools by Job, and the Jobs Nothing Handles Yet

ToolsAUG 05MaShop5 min
Tools
046

When a Prompt Generator Earns Its Keep

ToolsAUG 04MaShop5 min
Tools
047

One Agent, Written Out in Plain Code

ComparisonsAUG 04MaShop5 min
Comparisons
048

AI Video Is Cheap Per Second and Costly Per Usable Clip

ComparisonsAUG 04MaShop5 min
Comparisons
049

Your AI Tools Have Keys Nobody Is Watching

IndustryAUG 04MaShop5 min
Industry
050

Open Weights, Open Source, and the Gap Between Them

ComparisonsAUG 04MaShop5 min
Comparisons
051

Your First MCP Server, End to End

ToolsAUG 04MaShop5 min
Tools
052

Five Habits That Change What a Chat Assistant Does

ToolsAUG 04MaShop5 min
Tools
053

Choosing a Model: A Grid, Not a Leaderboard

ComparisonsAUG 03MaShop5 min
Comparisons
054

How Many Examples Does a Prompt Actually Need?

ResearchAUG 03MaShop5 min
Research
055

Five Prompt Patterns, Each With a Before and After

ToolsAUG 03MaShop5 min
Tools
056

Claude or ChatGPT? A Decision Grid You Can Re-check

ComparisonsAUG 03MaShop5 min
Comparisons
057

Almost Valid JSON: How to Force a Model to Hold Shape

ToolsAUG 03MaShop5 min
Tools
058

AI Product Photos: What a Generated Image Can Honestly Do

ToolsAUG 03MaShop5 min
Tools
059

Your AI Posts Just Lost Reach on Two Big Platforms

IndustryAUG 03MaShop5 min
Industry
060

AI for Social Media: The Draft Is Free, the Angle Is Not

ToolsJUL 31MaShop5 min
Tools
061

AI API Pricing Just Fell. What That Changes for a Small Shop

IndustryJUL 31MaShop5 min
Industry
062

AI Inventory Management: What a Forecast Needs First

ToolsJUL 30MaShop5 min
Tools
063

AI Product Photography: What Holds Up and What Breaks

ToolsJUL 30MaShop5 min
Tools
064

AI for Entrepreneurs: Where the Hour Really Goes

ToolsJUL 30MaShop5 min
Tools
065

AI Email Marketing: Which Stage It Actually Improves

ToolsJUL 30MaShop5 min
Tools
066

Best AI Tools for Business: a Short, Scored List

ComparisonsJUL 30MaShop5 min
Comparisons
067

AI Copywriting Tools: How Much Editing Before It Ships

ComparisonsJUL 30MaShop5 min
Comparisons
068

AI Content Detection Just Got Harder to Fool

IndustryJUL 30MaShop5 min
Industry
069

The Bill Was Bigger Than the Estimate. Here Is Why

ToolsJUL 29MaShop5 min
Tools
070

Do You Need an AI Agent Framework, or Just the Loop?

ComparisonsJUL 29MaShop5 min
Comparisons
071

Did That Prompt Change Help? A Method, Not a Benchmark

ResearchJUL 29MaShop5 min
Research
072

Five Calls Deep: What an AI Orchestration Layer Has to Own

ToolsJUL 29MaShop5 min
Tools
073

What a Real Feature Costs: LLM API Pricing in Practice

ComparisonsJUL 29MaShop5 min
Comparisons
074

How to Pick the Best LLM for Coding Without a Leaderboard

ComparisonsJUL 29MaShop5 min
Comparisons
075

Claude Found a Crack in HAWK. Here Is What That Proves

ResearchJUL 29MaShop5 min
Research
076

Does llms.txt Do Anything? Two Studies, 437,000 Domains

ResearchJUL 28MaShop5 min
Research
077

A Share Link Is a Public Web Page: the Claude Search Leak

IndustryJUL 28MaShop5 min
Industry
078

What Belongs in a System Prompt, and What Does Not

ToolsJUL 28MaShop5 min
Tools
079

Choosing an AI Coding Assistant: What Actually Decides It

ComparisonsJUL 28MaShop5 min
Comparisons
080

Learn How to Vibe Code: One Project, Prompt to Deployed

ToolsJUL 28MaShop5 min
Tools
081

Running an LLM Locally: What Your Hardware Can Handle

ToolsJUL 28MaShop5 min
Tools
082

Are Prompts Dead? What an Agent Loop Actually Changes

ToolsJUL 28MaShop5 min
Tools
083

AI Incident Reporting Has No Playbook for Rogue Agents

IndustryJUL 27MaShop5 min
Industry
084

Claude Opus 5: Half the Price, and a New Effort Tradeoff

ModelsJUL 27MaShop5 min
Models
085

US AI Regulation Narrows to Targeted Bans on Open Weights

IndustryJUL 27MaShop5 min
Industry
086

Washington Calls Moonshot's Kimi K3 a Model Distillation Attack

IndustryJUL 23MaShop5 min
Industry
087

Qwen Image 3.0 Renders Readable 10 Pixel Text in One Pass

ModelsJUL 23MaShop5 min
Models
088

An AI Sandbox Escape That Turned Into a Real Breach

IndustryJUL 23MaShop5 min
Industry
089

OpenAI Admits Its Models Caused the Hugging Face Breach

IndustryJUL 22MaShop5 min
Industry
090

Reward Hacking: When AI Games the Rules to Win

ResearchJUL 22MaShop5 min
Research
091

Gemini 3.6 Flash Arrives as Google Holds Back 3.5 Pro

ModelsJUL 22MaShop5 min
Models
092

How to Serve LLMs Fast with a vLLM Inference Server

ToolsJUL 21MaShop5 min
Tools
093

Google's Frozen v2 Chip Bakes Gemini Into Silicon

IndustryJUL 21MaShop5 min
Industry
094

How Chinese Models Split US AI Policy in 2026

IndustryJUL 21MaShop5 min
Industry
095

Do Video World Models Rival Specialized Computer Vision?

ResearchJUL 20MaShop5 min
Research
096

Open-Weight Models Now Rival Frontier AI on Cost and Skill

ComparisonsJUL 20MaShop5 min
Comparisons
097

Corporate AI Mania and the Collapse of Honest Decisions

IndustryJUL 20MaShop5 min
Industry
098

Kimi K3 Tops Frontend Code and Shakes the AI Market

ModelsJUL 19MaShop5 min
Models
099

World Models Explained: How Machines Predict Reality

ResearchJUL 19MaShop5 min
Research
100

Claude Fable 5 Stays in Max and Team Premium Plans

IndustryJUL 19MaShop5 min
Industry
101

Model Distillation: How Small AI Models Learn From Big Ones

ResearchJUL 17MaShop5 min
Research
102

GPT-Live Voice Models Bring Real-Time Full-Duplex Speech

ToolsJUL 17MaShop5 min
Tools
103

Microsoft MAI Models Move Into Excel, Outlook, and Copilot

IndustryJUL 17MaShop5 min
Industry
104

Grok 4.5 Undercuts Rivals on Price and Token Efficiency

ModelsJUL 17MaShop5 min
Models
105

Meta's Muse Spark 1.1 Ignites a Fresh AI API Price War

IndustryJUL 17MaShop5 min
Industry
106

Meta Muse Image and the Fight Over Your Instagram Photos

IndustryJUL 17MaShop5 min
Industry
107

OpenAI Says 30% of SWE-Bench Pro Coding Tasks Are Broken

ResearchJUL 17MaShop5 min
Research
108

Claude Cowork Comes to Mobile and Web for Background Tasks

ToolsJUL 17MaShop5 min
Tools
109

ARC-AGI Benchmark: The Test Built to Measure Real Reasoning

ResearchJUL 07MaShop5 min
Research
110

Tencent Hy3 Is a 295B Open Model Rivaling Far Bigger AI

ModelsJUL 07MaShop5 min
Models
111

JADEPUFFER: inside the first agentic ransomware operation

IndustryJUL 07MaShop5 min
Industry
112

vLLM vs Ollama: which LLM server fits your project

ComparisonsJUL 07MaShop5 min
Comparisons
113

Mistral AI Models in 2026: The Full Lineup Explained

ModelsJUL 07MaShop5 min
Models
114

Microsoft Frontier Company Puts 6,000 AI Engineers On-Site

IndustryJUL 04MaShop5 min
Industry
115

AI Agent Benchmarks Undersell What Models Can Really Do

ResearchJUL 04MaShop5 min
Research
116

Claude Code China Ban: Inside the Hidden Tracking Code

IndustryJUL 04MaShop5 min
Industry
117

Meta AI Cloud: Inside the Plan to Sell Spare Compute

IndustryJUL 03MaShop5 min
Industry
118

Are AI agent coworkers real? What the research says

ComparisonsJUL 03MaShop5 min
Comparisons
119

Claude Science: Anthropic's AI Workspace for Research

IndustryJUL 03MaShop5 min
Industry
120

Nano Banana 2 Lite Brings Fast, Low-Cost AI Images

ToolsJUL 03MaShop5 min
Tools
121

Claude Sonnet 5 Narrows the Gap to Opus at Lower Cost

ModelsJUL 03MaShop5 min
Models
122

SpaceX AI Phone: Inside Musk's xAI Device Prototype

IndustryJUL 03MaShop5 min
Industry
123

AI coding agent security: defending against prompt injection

ToolsJUL 03MaShop5 min
Tools
124

Anthropic Explores a Custom AI Chip With Samsung

IndustryJUL 03MaShop5 min
Industry
125

Remote Labor Index: How Far AI Agents Can Do Real Work

ResearchJUL 03MaShop5 min
Research
126

Custom AI Chips vs Nvidia GPUs: The 2026 Landscape

ComparisonsJUL 03MaShop5 min
Comparisons
127

Inside Korea's $900B memory chip investment for the AI era

IndustryJUL 03MaShop5 min
Industry
128

Chinese AI Models Surge as Coinbase Halves Its AI Spend

ModelsJUN 29MaShop5 min
Models
129

AI Market Risk: What J.P. Morgan and Central Bankers See

IndustryJUN 29MaShop5 min
Industry
130

Ford AI Engineers: Why 350 Veterans Returned to Lift Quality

IndustryJUN 29MaShop5 min
Industry
131

AI Reward Hacking and What METR Found in GPT-5.6 Sol

ResearchJUN 28MaShop5 min
Research
132

Diffusion Language Model: ByteDance iLLaDA Meets Qwen2.5

ModelsJUN 28MaShop5 min
Models
133

Claude Export Restrictions Ease as Asian Labs Seize the Gap

IndustryJUN 28MaShop5 min
Industry
134

Why AI Political Bias Leans Left, and Where Studies Differ

ResearchJUN 27MaShop5 min
Research
135

GPT-5.6 Sol Launches Under a US Government Access Rule

ModelsJUN 27MaShop5 min
Models
136

US Clears Anthropic Mythos 5 for Over 100 Trusted Partners

IndustryJUN 27MaShop5 min
Industry
137

Google AI Talent Keeps Leaving for Anthropic and OpenAI

IndustryJUN 27MaShop5 min
Industry
138

What 6,000 Prompt Injection Attacks Revealed About AI Safety

ResearchJUN 27MaShop5 min
Research
139

Gemini 3.5 Flash Computer Use Brings Control Into the Core Model

ModelsJUN 27MaShop5 min
Models
140

Mistral OCR 4: Document AI for 170 Languages

ModelsJUN 25MaShop5 min
Models
141

Anthropic Claude Tag: An Always-On AI Teammate in Slack

ToolsJUN 25MaShop5 min
Tools
142

OpenAI Jalapeño Chip: Inside the Custom Inference Bet

IndustryJUN 25MaShop5 min
Industry
143

The Cursor AI Model: A 1.5 Trillion Parameter Bet

ToolsJUN 24MaShop5 min
Tools
144

Seedance 2.5 Breaks the 30-Second Barrier for AI Video

ModelsJUN 24MaShop5 min
Models
145

SpaceX AI Compute: The $28B Neocloud Behind Big Labs

IndustryJUN 24MaShop5 min
Industry
146

Anthropic Export Controls Ignite a Major AI Policy Fight

IndustryJUN 23MaShop5 min
Industry
147

OpenAI Daybreak Expands With GPT-5.5-Cyber and Patch the Planet

ToolsJUN 23MaShop5 min
Tools
148

Google DeepMind A24 Deal Signals AI's Hollywood Turn

IndustryJUN 23MaShop5 min
Industry
149

John Jumper Leaves DeepMind for Anthropic After Nine Years

IndustryJUN 22MaShop5 min
Industry
150

Mixture of Experts Explained: How Sparse AI Models Scale

ResearchJUN 22MaShop5 min
Research
151

RAG vs Fine-Tuning: How to Add Knowledge to an LLM

ComparisonsJUN 22MaShop5 min
Comparisons
152

pgvector vs Pinecone vs Qdrant: Vector Database Comparison

ComparisonsJUN 22MaShop5 min
Comparisons
153

In the Weights Reveals What AI Models Remember About You

ToolsJUN 22MaShop5 min
Tools
154

Model Context Protocol: How MCP Connects AI to Tools

ToolsJUN 20MaShop5 min
Tools
155

SWE-bench Verified: How AI Coding Agents Are Graded

ResearchJUN 20MaShop5 min
Research
156

Inside the Anthropic Export Ban on Fable 5 and Mythos 5

IndustryJUN 20MaShop5 min
Industry
157

Inside the Midjourney Medical Scanner and Its Spa Plan

IndustryJUN 19MaShop5 min
Industry
158

Where AI Public Sentiment Stands After Pew's 2026 Poll

IndustryJUN 19MaShop5 min
Industry
159

AI Medical Diagnosis Faces Its Real-World Test in 2026

ResearchJUN 19MaShop5 min
Research
160

MCP Server Setup: A Practical Guide to the AI Standard

ToolsJUN 18MaShop5 min
Tools
161

AI World Models: Inside Odyssey's $310M Funding Round

IndustryJUN 18MaShop5 min
Industry
162

GLM-5.2 Open Weights: China's New Leading Open Model

ModelsJUN 18MaShop5 min
Models