AI & Agents
Claude Opus 5: The Effort Ladder and a Million Tokens
Thinking is on by default, the context window is a million tokens, and effort is now the main dial you turn. What that changes for a delivery team’s stack.
Read
Notes from the work
What we have learned building and rescuing products for SaaS companies and startups — architecture, AI, security, delivery. No listicles, no reheated press releases.
AI & Agents
MCP Goes Stateless: What the 2026-07-28 Spec Changes
Session IDs and the initialize handshake are gone. Any request can land on any instance, which removes sticky routing and changes how you deploy an MCP server.
Read
AI & Agents
OpenAI’s Agent Stack After AgentKit
Agent Builder and the Evals platform shut down on 30 November 2026. Choosing between the Agents SDK, Workspace Agents and the Responses API, and what to do about your evals.
Read
AI & Agents
Using Jasper AI: Where It Fits in a Content Workflow
A marketing AI tool is only as good as the process around it. What Jasper is genuinely useful for, what it produces badly, and how to keep the output from sounding like everyone else.
Read
AI & Agents
Using Midjourney for Quality Image Generation
Getting one striking image is easy. Getting forty that look like they came from the same studio is the actual skill, and it comes down to constraint, not cleverness.
Read
AI & Agents
Python: LangChain and LangGraph, Current State
LangChain got a reputation for abstraction it has partly earned and partly outgrown. Where the split with LangGraph leaves things, and when either is the right dependency.
Read
AI & Agents
Generative AI, RAG and Agentic Frameworks
Three overlapping terms that describe three different amounts of risk. What separates a prompt from a retrieval system from an agent, and which one your problem actually needs.
Read
Let's talk
Have a product in mind?
Tell us about your web, mobile, or agentic AI project — we'll get back with a tailored proposal, usually within a business day.