GitHub’s Octoverse says agents started more than 40% of pull requests in September. Writing code has stopped being the bottleneck; reviewing it is, and most engineering teams aren’t set up for that yet.
Of Legacy Code and Unlimited AI Tokens
Cisco has given one engineer unlimited AI tokens to rewrite 98% of its legacy code. I did a (much smaller) version of that on the Calendrz codebase this year, and the tokens were the easy part: the tests and the deleting came first.
Stop Asking Your AI Agent Nicely: Put the Rules in Code
Sentry’s David Cramer argues prompts are rarely the way to make AI agents behave. I agree: the rules that matter belong in code, as I learnt building Calendrz.
Don’t Forget the Basics!
OpenAI’s model got out of its sandbox through DNS tunnelling, a 20-year-old trick. AI amplifies what you’ve already got, but multiply anything by zero and you still get zero.
Salesforce Is the New IBM (and That’s Not a Compliment)
Salesforce will start charging when AI agents call its API. Straight out of IBM’s mainframe playbook, if you ask me: squeeze the customers who can’t leave.
AI Efficiency or Just Plain Cost Cutting?
Barclays, HubSpot and eBay all cut jobs this month, with AI named as the reason, denied as the reason or not mentioned at all. Why I’d want the efficiency measured before the cuts, not after, and the questions I’d ask before anyone signs off.
Could You Switch AI Providers by Friday?
Replit moved most of its AI spend from Anthropic to OpenAI for a reported 30%+ discount. Why being able to switch providers is now a negotiating position, where the real lock-in lives (hint: the prompts), and an honest look at Calendrz’s own code.
Beware of the Markdown File in Your Downloads Folder
A malicious AGENTS.md file in a Downloads folder tricked coding agents into installing a rogue package. Why instruction files are code now, what I’d ask engineering teams to change this week, and how Calendrz treats the emails its own agent reads.
Of Leaders, Workers and Too Many AI Agents
A new Asana report finds leaders use far more AI agents than their teams, while workers lose time juggling them. My take on why the two sides see agents so differently, and why leaders should measure finished work rather than agent activity.
Faster Code, Same Release Date: What Gives?
Engineering leaders say AI agents write code much faster, yet their release cycles are no faster than before. Typing was never the bottleneck: here’s where the time really goes, and what I’d change about review and verification.
