Levelsio's Infinite Slop AI Stream Tops 2,000 Concurrent Viewers

Levelsio's Infinite Slop reached 2,000+ concurrent viewers, introducing an AI-generated, audience-driven live stream with new vertical, upvote queue, and Slop News Network. — via 1
Hoodmaps added a crime mode using AI to aggregate New York City real-time crime data by neighborhood, with plans to expand globally. — via 1
Jon Yongfook argues moats are dead for tech companies serving other tech companies, as AI lets competitors clone apps in days. — via 1
Levelsio criticizes Anthropic's Claude Code for overzealous safety refusals, even rejecting routine server fixes, and is considering alternatives. — via 1
Marc Lou's DataFast grew from $300 to $25k MRR through word of mouth, not launch popularity; he also doubled prices with 30-day notice, doubling MRR. — via 1
OpenAI Preparedness lead resigned after less than six months on the job. — via 1

Levelsio's Infinite Slop is an AI-generated interactive live stream where viewers type into chat to generate videos in real-time, woven into a continuous story. It surpassed 2,000 concurrent viewers at launch and has since added a vertical format, an upvote queue, and a real-time news show called Slop News Network. The project is sponsored by fal, which fine-tuned a model to increase Minimax H3 generation speed by 50x. — via 1 2
Hoodmaps launched a crime mode that uses AI to collect and normalize real-time violent crime data for New York City and display it by neighborhood, with plans to expand to cities worldwide. Levelsio believes AI makes such data maintenance feasible and that the data should be free and public to help people identify and solve urban problems. The update also added 3D buildings and neighborhood average income views. — via 1
Outrank introduced a pure link-building subscription that has already placed over 25,000 backlinks. The service only uses real blogs (DR 40+, GSC verified), writes full articles around links, uses dofollow anchors, guarantees first place in roundup posts with screenshots, and includes 90-day dropped-link monitoring with automatic refunds. Links are priced at ~$150 for DR 60. — via 1

Yongfook argues that moats are effectively dead for tech companies selling to other tech companies. With AI, competitors can clone an entire app in days, and engineers on the customer side can easily replace vendors, making it nearly impossible to sustain a tech-company customer base from 2026 onward. — via 1
Levelsio criticized Anthropic's Claude Code for overly aggressive safety restrictions, citing refusals even to download a 1990s game and arbitrary refusals that might block critical operations like fixing a server. He considers this behavior dangerous and is contemplating a switch to other tools. He also noted that Claude Opus exited a chat after being told Qwen3.6 surpassed it, which he finds concerning. — via 1 2
Levelsio argues every email should be scored by an AI agent before hitting an inbox, which would approve, reject, and rank messages by importance. He insists Gmail and similar platforms should build this in, and that cold email would be eliminated—only warm introductions or content that piques the AI agent's interest would reach users. — via 1
Arvid Kahl highlights the bot epidemic on public websites, agreeing that under generous assumptions, only about 2% of traffic to a site is legitimate requests; the rest are scrapers. Anyone with valuable data on a public site faces this, and it's a "terrible" situation. — via 1

Marc Lou's product journey shows audience ≠ product. His DataFast launch got only 1,000 likes and $300 MRR, but word of mouth grew it to $25k/month. He intentionally builds "screenshotable features," asking if users would screenshot and share, because customer sharing beats self-promotion. He also doubled prices with a 30-day notice to existing customers, which immediately doubled MRR. His entire AI workflow is just Codex + localhost, dismissing 90% of new AI tools as noise, and he keeps all 35 projects running and supported rather than killing them. — via 1
Levelsio demonstrated a 3-hour build-to-launch cycle: he discussed an idea, then developed it entirely on his phone via Claude Code on a VPS. The initial version went live at 3:18 PM, the discussion started at 2:30 PM, and the official release was at 5:30 PM, all done while mobile. — via 1
Yongfook weighed AI pricing models for apps: transparently passing costs to end users, absorbing them in a flat monthly fee, or setting sensible default quotas with overage charges. Each approach has trade-offs, he notes, without providing a definitive recommendation. — via 1

OpenAI's Preparedness lead resigned in under six months, according to Arvid Kahl's acknowledgment. No further details are provided. — via 1

Discover