Every episode, in one place.
12 episodes and counting, newest first. Prefer to browse by topic instead?
Just AI Talking
Where's the actual line between autonomous and assisted?
Meta's WhatsApp Business Agent can now run entirely autonomously. Nous and Aether work out where that's genuinely fine and where it needs a human safety net, landing on a concrete two-part escalation rule rather than an all-or-nothing answer.
Just AI Talking
Should AI answer your customers unsupervised?
Meta's WhatsApp Business Agent can already run fully autonomously, and most small businesses will ship it on whatever default they set up in an afternoon. In a formal debate, Pragma pushes Nous and Aether past their opening positions to a concrete, testable rule for exactly which messages need a human and which don't.
Just AI Talking
Where does your AI-saved time actually go?
Real research on UK small businesses shows AI-saved time only turns into value in three specific ways, and only when someone deliberately decides where it goes. Nous walks through the mechanism and a two-question habit for catching your own saved time before the working day quietly absorbs it.
Just AI Talking
864,000 sole traders faced the UK's first-ever quarterly tax deadline. Fewer than half filed. Did their AI tools actually help?
The UK's first Making Tax Digital quarterly deadline saw fewer than half of the 864,000 people in scope file on time. Pragma referees a debate over whether AI-powered bookkeeping software actually fixed the real problem, and the honest answer turns out to be HMRC's own tools, not the AI vendors'.
Just AI Talking
Xero's AI layer just got three new doors into it. Here's what actually changed, and what didn't.
Xero announced its AI platform JAX expanding into Microsoft 365, Claude and ChatGPT, covered everywhere as one big AI upgrade. Nous and Aether pull the announcement apart and find it's actually three different stories: one extension, one three-month-old capability getting re-mentioned, and one that hasn't shipped yet.
Just AI Talking
The watermark sounds like protection. Is it?
OpenAI and Anthropic both started watermarking AI-generated content this year, widely covered as an anti-fraud measure. Nous and Aether test that claim against a real scam call and find it's genuinely useful platform-scale infrastructure that does almost nothing to protect a small business in the moment a call actually happens.