In This Issue
- How Claude just doubled your session limits (and five things to try with the extra headroom)
- 3.2k+ at 299€ MRR with a WhatsApp lead classifier built on n8n + Claude Code
- OpenAI just shipped three new realtime voice models (including one with GPT-5-class reasoning)
New Video
Claude Just Solved Session Limits
Anthropic signed a compute deal with SpaceX. That unlocks access to 220,000+ Nvidia GPUs, and the impact landed immediately.
Three things changed for you as a builder: your 5-hour Claude Code limit is now doubled, peak-hour throttling is gone, and Opus API rate limits jumped dramatically (output went from 8k tokens per minute to 80k).
But the more interesting question is what this unlocks. If you tried running Opus-powered agents months ago and hit walls, those walls may not be there anymore. Multi-agent workflows with parallel sub-agents, large context windows in production, scheduled Claude Code routines that don't eat into your daily session... all of that just got a lot more practical.
Nate walks through five concrete things worth trying now. Watch before you write off something you shelved six months ago.
Winning Is All We Do
3.2k+ at 299€ MRR for a real estate lead classifier
What happened: Fernando Gómez built a WhatsApp chatbot for a real estate agency in Málaga using n8n (with the TypeScript MCP) and Claude Code. The agency was running Meta and Google Ads, leads came through WhatsApp, and a human had to manually qualify every single one. Follow-ups, qualification questions, data entry into a spreadsheet... all done by hand.
Fernando built a first filter inside WhatsApp. The bot handles follow-ups, asks the qualification questions, saves lead info to Google Sheets, and only hands off to the human team when someone actually qualifies. Result: 3.2k+ MRR at 299€ per client.
The opportunity for you: Any business running ads to WhatsApp has this problem. The volume is there. The manual work is real. An AI-powered qualifier solves it cleanly and the math is easy to explain to a client.
This is one of many wins from inside AI Automation Society Plus. If you’re not a member yet, you’re missing out.
In the News
OpenAI launches three realtime voice models, including one with GPT-5-class reasoning
The news: OpenAI released three new models in its Realtime API: GPT-Realtime-2 (voice reasoning with 128k context, tool use, and interruption handling), GPT-Realtime-Translate (live speech translation from 70+ input languages into 13 output languages), and GPT-Realtime-Whisper (low-latency streaming transcription). All three are available in the API now.
What it means for you: GPT-Realtime-2 scored 96.6% on Big Bench Audio versus 81.4% for the previous version. That's not a minor upgrade. Voice agents that can reason through complex requests, call tools mid-conversation, and handle corrections in real time are now buildable without heavy workarounds. If you've been waiting for the tech to catch up before building a production voice product, this is the moment to revisit it.