Claude used to grade first dates

A woman is recording her first dates with an app called Granola, sending the transcripts to Claude, and asking the AI to grade how well the evening went. The Wall Street Journal profiled her, and the story has been making rounds on social media as a sharp example of AI creeping into corners of life nobody saw coming.

This is a single-source anecdote, not a trend study, but it’s a useful signal. People are already comfortable enough with voice transcription and LLMs to hand over something as personal as a date recap and ask for feedback. The barrier to entry is gone. You open an app, hit record, paste the text into Claude, and get a structured debrief. No technical skill required.

For business owners, the takeaway is not about dating. It’s that your customers and employees are doing this kind of thing already, in contexts you might not expect. They’re recording meetings, sales calls, customer complaints, and performance reviews, then feeding them to an LLM for analysis or advice. Some of that is harmless. Some of it creates risk if sensitive information leaves your systems without guardrails.

The smarter move is to build this capability inside your operation with controls you set. An AI command centre can transcribe, analyse, and surface insights from internal conversations without data leaking to consumer apps. You decide what gets recorded, who sees the output, and how the analysis runs. If your people want AI feedback loops, you want those loops running on infrastructure you control, not on someone’s personal ChatGPT account after a client call. The technology is simple now. The governance part is what separates companies that use AI well from companies that discover too late where their data went.