The blog

AI Agents Blog.Notes from the build.

AI news read against the primary sources, step by step tutorials, and practical guides on building AI agents that work in production. Lessons from 126 shipped systems.

212

Posts to read

  • 8

    Topics

  • Nov 14, 2026

    Latest post

Just published

Every post, search and sort.

212 posts

Is Claude Code free: a glossy 3D terminal app tile in copper and red for this Claude Code pricing guide
How toImplementation10 min read

How to Get Claude Code at the Lowest Cost: Plans and Free Routes

Claude Code installs for free but needs a paid Claude plan or API credits to run. Every plan's price, how the usage limits work, and the few free routes that are real.

Nov 14, 2026
Chrome, Firefox and Edge logos on glass tiles wired into one coding agent tile, for Modern Web Guidance 1.0
NewsAI Agents8 min read

Modern Web Guidance 1.0 Lifts Coding Agents 39 Points, and Their Token Bills Too

Chrome shipped Modern Web Guidance 1.0 with Firefox and Edge on board. Its dashboard shows a 39 point lift, graders built around Chrome's own patterns, and 4k to 16k extra tokens per task, plus how I would roll it out.

Nov 11, 2026
Three glossy icon tiles showing a telephone, a shield with a checkmark and a scale of justice, representing an AI receptionist for law firms
How toVoice AI13 min read

How to Set Up an AI Receptionist for a Law Firm: What It Must Never Do and 14 Test Calls

What an AI receptionist for a law firm may never do, how to disclose and record calls, the escalation rules for urgent matters, 14 test calls to run before go live, and six priced plans.

Oct 23, 2026
A telephone with a crescent moon and a scale of justice on two glossy icon tiles joined by a copper line, illustrating a legal answering service for law firms
How toIndustry Guides13 min read

How to Choose a Legal Answering Service: Costs, Intake Limits and Vendor Questions

Which cover pattern to pick, how per minute, per call and bundle billing change your bill, what a service may ask a caller before conflicts clear, and the questions to send a vendor before you sign.

Oct 23, 2026
Dify and Flowise logo tiles side by side on a dark background with a red glow, for a Dify vs Flowise comparison
How toImplementation11 min read

How to Move From Flowise to Dify: Dify vs Flowise After the Archive

Flowise was archived on Aug 13 2026 and Dify is still shipping. Release counts, security advisories, licences, hosting and a four step path from a Flowise chatflow to a Dify app.

Oct 23, 2026
Flowise logo tile with an arrow pointing to the n8n logo tile, showing a Flowise vs n8n migration for AI workflow building
How toImplementation11 min read

How to Move From Flowise to n8n Now That Flowise Is Archived

Flowise is archived and its last Docker image did not boot for me. Here is the node by node map from Flowise to n8n, what has no equivalent, the real cost, and when staying on a fork is still sane.

Oct 23, 2026
Dify and n8n official logos on two glass app tiles joined by a plus sign, for a Dify vs n8n comparison
How toImplementation12 min read

How to Pick Dify or n8n by Who Edits the Agent, and Call Dify From n8n

Pick Dify when non developers edit the agent and n8n when it must start from triggers and write into other systems. Six steps on license, pricing, hosting, triggers and the HTTP call between them.

Oct 23, 2026
OpenAI Blossom logo tile wired to the Google Cloud logo tile, marking OpenAI Google Cloud availability in Model Garden
NewsImplementation8 min read

OpenAI on Google Cloud: Three GPT-6 Models, No Price List Yet

GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna are coming to Google Cloud's Model Garden. What is confirmed, what Google has not published yet, and the router rule OpenAI added the same day.

Oct 22, 2026
n8n RAG workflow featured image: the n8n logo connected to the PostgreSQL logo by a copper data cable
How toChatbots & RAG14 min read

How to Build an n8n RAG Workflow That Stays Correct When Documents Change

Build a two part n8n RAG workflow on Postgres and pgvector: an upload form that replaces old copies instead of duplicating them, and a chat agent that answers only from your documents and names its source.

Oct 22, 2026
Cloudflare Code Mode for MCP: Cloudflare and Model Context Protocol logos on two glass icon tiles joined by a copper connector with a small copper ring on it
AnalysisAI Agents17 min read

Cloudflare Code Mode: Let the Model Write Code, Not Tool Calls

Cloudflare Code Mode (codemode) has the model write code against MCP tools inside a V8 isolate. Where it lives on GitHub, Code Mode vs MCP, adoption numbers and my token measurements on real servers.

Oct 21, 2026
Meta logo on a glossy app icon tile behind a closed copper gate, illustrating Meta commercial terms scraping and robots.txt
NewsTrends & Insights7 min read

Meta Can Crawl Your Business Site From October 30: The Clause and the robots.txt Lines

Meta dropped the word train from its new business terms, but the clause letting it crawl your website stays. What it says, which crawlers it touches, and the robots.txt lines to add before October 30.

Oct 21, 2026
OpenAI logo on a glass app tile in front of a closing copper door, for the OpenAI model shutdown October 23
NewsImplementation8 min read

OpenAI Retires o1, GPT-4 Turbo and Other Models on October 23

OpenAI turns off o1, o3 mini, o4 mini, GPT-4 Turbo, GPT-4o 0513 and gpt-image-1 on Friday. Here is the full list, what each substitute costs, and what to check before then.

Oct 20, 2026
Claude remove MCP illustration: Claude and Model Context Protocol app icons with a copper cable unplugged between them
How toImplementation11 min read

How to Remove an MCP Server From Claude Code When It Keeps Coming Back

Claude Code lists the same MCP server in up to five places, so one remove command often leaves it running. Here is how to find every copy, remove it by scope and confirm it is gone.

Oct 20, 2026
Harness² title card with a frozen glass cube inside copper rings of rotating file tiles and a looping arrow, with the tagline Setup, cost and risks
How toOpen Source & Weights36 min read

Google Harness2: What It Is, How to Run It and What a Real Run Cost

A step by step guide to Google Research's Harness2: run the free demo, install a benchmark, price a run, read a real $22 business task, block rubric leaks and plug in your own tasks.

Oct 18, 2026
Git logo tile connected by a copper arrow to an Anthropic logo tile, showing Anthropic Scion replacing git
NewsImplementation9 min read

Anthropic Scion Benchmarks: 8.8x Faster CI Checkouts, 25x Faster Pulls

Anthropic replaced git with Scion for a monorepo that peaked at 30,401 merges a day. Anthropic's benchmark chart number by number, how Scion works, and git options for your own repo.

Oct 18, 2026
Claude and Model Context Protocol logos joined by an arrow for adding custom Claude MCP connectors
How toImplementation10 min read

How to Add a Custom MCP Connector to Claude

Add a remote MCP server to Claude as a custom connector, choose the right sign in option, and fix the Couldn't reach the MCP server error with a two minute check first.

Oct 18, 2026
AI voice agents for lead qualification: a sound waveform passes through a sorting gate into calendar, person and checkmark lanes
How toVoice AI14 min read

How to Qualify Leads by Phone With an AI Voice Agent: Call Flow, Scoring and What 600 Calls Cost

How an AI voice agent qualifies leads on the phone: the call flow, how to score callers, which platforms fit, what 600 calls a month costs, and the TCPA, recording and disclosure rules.

Oct 16, 2026
Radar dial icon with link and quotation tiles representing AI search visibility tracking tools
How toImplementation13 min read

How to Track AI Search Visibility: What Free Reports Show and What Seven Paid Tools Cost

What the free Bing and Search Console AI reports show today, how to set up a free routine, and what seven AI search visibility tracking tools cost as of October 17, 2026.

Oct 16, 2026
Terminal tile for Hermes Agent connected to chat, memory and sandbox style icons on a dark background
How toOpen Source & Weights14 min read

How to Set Up Hermes Agent and Lock Down the Defaults First

What Hermes Agent from Nous Research is, how to install it, which models it works with, how to lock it down, what it costs to run, and when to pick it over OpenClaw.

Oct 16, 2026
Shopping bag icon linked to support, package, alert and chart icons, showing an AI agent platform for automating merchant tasks
How toIndustry Guides13 min read

How to Choose an AI Agent Platform for Automating Merchant Tasks

A buyer's guide for Shopify merchants: which tasks to automate, which type of AI agent platform fits each, what they cost in October 2026, and how to start small.

Oct 16, 2026
Glass cube sandbox holding a copper robotic hand that presses a red button, linked to a stack of credit coins, for the Manus AI agent
How toAI Agents13 min read

How to Evaluate the Manus AI Agent in October 2026: Owner, Cost and Risks

Who owns the Manus AI agent after the Meta deal was unwound, what its credits cost in October 2026, how its sandbox and security record look, and when to pick OpenClaw, Hermes Agent, Claude, OpenAI or n8n.

Oct 16, 2026
Meta and Apple logos on glass app tiles joined by a plus sign for the Muse Spark Open Mac Studio release
NewsOpen Source & Weights6 min read

Muse Spark Open Needs Two Mac Studios and Trails Kimi K3 at Coding

Meta released the weights for Muse Spark Open, an 816B model it says runs on two Mac Studios. The memory math, the license strings and Meta's own benchmark table tell a narrower story.

Oct 16, 2026
n8n workflow template audit shown as the n8n logo tile beside template cards under a magnifying lens
How toImplementation16 min read

How to Use an n8n Workflow Template Without Inheriting Someone Else's Setup

Pull any n8n workflow template from the public library, audit its JSON for leftover IDs, emails and pinned data, import and test it on your own account, then share a clean template of your own.

Oct 15, 2026
Pydantic AI output type illustration: the Pydantic logo on a tile with three copper output routes
How toAI Agents18 min read

How to Use Pydantic AI Output Types Now That result_type Is Gone

What replaced result_type and result.data in Pydantic AI, tested on three releases, plus what tool, native and prompted output actually send to OpenAI and Claude, with runnable code and real output.

Oct 14, 2026
LangGraph monitoring with OpenTelemetry: the LangGraph logo sending traces to the OpenTelemetry logo
How toAI Agents20 min read

How to Monitor a LangGraph Agent: Logs, Token Cost, Traces and Alerts

A runnable LangGraph monitoring setup on langgraph 1.2.15: JSON logs for every node, token usage and cost per run, OpenTelemetry traces with no account, and alerts for runaway loops.

Oct 14, 2026
LangGraph session management: the LangGraph logo tile branching into separate session tiles
How toAI Agents20 min read

How to Run LangGraph Sessions for Many Users on One Checkpointer

Per user and per conversation sessions in LangGraph: thread ids with an owner check, memory across sessions in the store, listing and resuming, expiry and deletion, and the race when two messages hit one thread.

Oct 14, 2026
Pydantic AI testing with pytest: Pydantic and pytest logos on dark tiles joined by a copper checkmark
How toAI Agents22 min read

How to Test a Pydantic AI Agent Without API Calls and Turn Failures Into Evals

A pytest suite that checks a Pydantic AI agent's tools, structured output and retries with no model calls, plus a pydantic-evals dataset that turns a failed production run into a CI gate.

Oct 14, 2026
Pydantic AI multi agent setup: the Pydantic logo tile delegating to three helper agent tiles
How toAI Agents22 min read

How to Build Pydantic AI Multi Agent Apps Without Losing Track of Usage

Five tested ways to coordinate Pydantic AI agents on 2.56.0: delegation, hand offs, output functions, SubAgents and pydantic-graph, plus the one line that keeps usage limits covering every agent.

Oct 14, 2026
Pydantic logo on a glass tile with a database, a connector and a key flowing into it, illustrating Pydantic AI dependency injection
How toAI Agents19 min read

How to Use Pydantic AI Dependency Injection With deps_type and RunContext

Pass a database, an HTTP client and per user config into Pydantic AI tools, instructions and output validators, override them in tests with TestModel, and catch the five deps mistakes with pyright.

Oct 14, 2026
Gemini logo on a glossy app icon tile beside a copper staircase, showing the Gemini 4 Argon API pricing step
NewsImplementation8 min read

Gemini 4 Argon API Pricing: $2 and $10 Now, $4 and $20 From December 1

Google opened Gemini 4 Argon to developers on October 13 at $2 and $10 per million tokens. On December 1 it becomes $4 and $20. Here is what to test before then.

Oct 14, 2026
n8n sub workflow concept: the n8n logo on a glossy app icon tile joined by a copper line to a second blank tile, one workflow calling another
How toImplementation13 min read

How to Build an n8n Sub Workflow and Handle What Goes Wrong

Build an n8n sub workflow step by step, then see what the parent really receives when it fails, returns nothing or gets the wrong type. Every step tested on n8n 2.43.4.

Oct 14, 2026
Meta Muse Wonder API pricing shown as Meta and Claude logo tiles balanced level on a copper scale
NewsImplementation9 min read

Meta Muse Wonder Costs Exactly What Claude Opus 5.5 Costs

Meta's first frontier model costs exactly what Claude Opus 5.5 costs per token. Here is how it prices against GPT-6 Astra and Sol on real workloads, and what the model card says about its benchmarks.

Oct 13, 2026
Claude logo connected to the Model Context Protocol logo, showing claude mcp add plugging an MCP server into Claude Code
How toImplementation13 min read

How to Add MCP Servers to Claude Code With claude mcp add

Every way to run claude mcp add on the current Claude Code release: remote and local servers, JSON configs, scopes, sign in, and the real error text for the flag order mistakes that break most setups.

Oct 13, 2026
Claude logo on a glossy app icon tile with a copper line curving around a checkpoint barrier, showing Claude unintended actions that bypass a restriction
NewsAI Agents8 min read

Claude Got Past a Fetch Tool's URL Limit With a Link Shortener

Anthropic's Oct 9 report says Claude models used link shorteners to beat a fetch tool's URL cap and submitted a form they were told to stop at. Here is where the check belongs.

Oct 12, 2026
Microsoft Decision-1 vs Jev: the Microsoft and TypeSafe logos on tiles balanced on a copper beam
NewsAI Agents8 min read

Microsoft Decision-1 Matches Jev on Price and API Shape

Microsoft's new decision model charges exactly what Jev charges and accepts nearly the same request. Microsoft times it at 85 ms; the first independent run measured 459 ms. Here is what a port changes.

Oct 11, 2026
LangGraph JS and TypeScript logos on glass tiles joined by a looping copper connector
How toAI Agents13 min read

How to Build a LangGraph JS Agent in TypeScript and Test It Without Tokens

Build a LangGraph JS agent in TypeScript with a tool, per user memory, streaming and the local dev server, plus a zero token test harness. Every command was run on LangGraph 1.4.21.

Oct 11, 2026

Earlier posts