JetThoughts has been building software since 2008. Book a call, or write to sales@jetthoughts.com. You'll leave the call with a written one-page scope within 2 working days.
A 30-minute call; pick a time.
Browse by topic
For founders ยท For engineersLatest posts
RSS
An AI Agent Deleted a TV Show's Whole Season
Five weeks before its premiere, The Challenge Sverige lost its edited season to an AI agent on the production server. Here's what founders can take from it.
· JetThoughts
Anthropic: AI Worms Are Months Away
Anthropic's red team showed small AI models hopping between $55 Android TV boxes. No rogue worm exists yet. Here's what the experiment does and doesn't prove.
· JetThoughts
MCP's Command Execution Flaw Is 'By Design'
OX Security found 200,000 MCP servers exposed to command execution, and Anthropic says the behavior is intended. Here's what to check if your team runs MCP.
· JetThoughts
Stack Overflow 2026: Devs Verify AI Code
Claude Code passed GitHub Copilot in Stack Overflow's 2026 survey, and 76.5% of developers run AI output locally before using it. Only 9.6% use it as-is.
· JetThoughts
Should You Let Them Build It in Rails?
You cannot judge a framework you cannot read, and comparing frameworks across ecosystems mostly produces noise. Here is what a non-technical founder can actually check when someone proposes Ruby on Rails.
· Paul Keen
Autonomous Agent Teams Are Real. Measured Success Runs 13% to 59%.
Six multi-agent frameworks, measured across coding and reasoning benchmarks. Then the pattern both the loudest sceptic and the loudest proponent ended up agreeing on: many agents may think, one agent writes.
· Paul Keen
Have an App Idea? Here's Where to Start
Turn your app idea into a clear concept and a small prototype. A practical guide for non-technical founders, with rough sketches and AI prompts you can copy.
· Paul Keen
"Generate Then Verify" Moved Your Work. It Didn't Remove It.
81% of engineering teams report spending more time in code review since adopting AI tools. The advice assumes checking is the cheap half. Measurement says that is where the hours went.
· Paul Keen
Your AI Reviewer Approved It. Did It Read Its Own Report?
Our AI code reviewer found a SQL injection, described it accurately, and set the verdict to approve in the same response. Here is what we changed.
Your Dev Shop Says the Code Was Reviewed. By What?
Copilot now takes part in more than one in five code reviews on GitHub. The research says AI reviewers earn their keep - and that the tuning which makes them usable suppressed about a fifth of real vulnerabilities on a standard benchmark.
· Paul KeenTell us what you need built