My OpenAI dot read my email before I gave it any job — then I ran it against my Claude Code routine

added Oct 6, 2026

I set up OpenAI's new dot, and before I gave it a single job, it had already gone through my email. So I gave an OpenAI dot and my Claude Code routine the exact same 6 AM job and compared what came back. In this video, I show my real dot (Scout) and the Claude Code routine I've run every morning since August, side by side. You'll see what OpenAI Dots actually are, how they differ from Claude Code routines on five things (who starts the work, memory, where it works, price and control), and the one rule for any AI agent that works while you sleep. Free one-page map + notes: https://buildalone.co/dots 00:00 It read my email before I asked 00:31 What an OpenAI dot actually is 01:11 Same job, 6 AM: Scout vs my Claude routine 02:04 ① Who starts the work 02:22 ② Memory (and the repeat problem) 03:04 ③ Where it works 03:19 ④ Price: $100 vs $20 03:45 ⑤ Control 04:11 Which one should you pick? 04:33 The warning: give it the smallest key 05:06 Recap Build Alone: practical AI for people building on their own. https://buildalone.co/ Start here: 13 AI terms in plain English → https://youtu.be/1eVqXjZZjKo #OpenAIDots #ClaudeCode #AIAgents

Builder channel Build Alone set up his own OpenAI dot 'Scout' and ran it side-by-side against the Claude Code morning routine he's run since August. Concrete result shown on video: before receiving a single job, the dot had already gone through his email on its own. He then gave both agents the exact same 6 AM job and compared the outcomes across five dimensions (who starts the work, memory, where it works, price: $100 vs $20, control), with his real dot visible on screen. First-hand test with concrete autonomous-email behavior demonstrated, not a hypothetical. Real channel name: Build Alone.