summree
My AI Agents Clipped This Stream While I Slept
← Claude
Creator Magic

My AI Agents Clipped This Stream While I Slept

⏱ 64 min video · 3 min read13 Jul 2026Worth watching
TL;DR
Mike Russell demos 'Gaia Clipper,' a custom AI agent system he built that automatically clips highlights from his live streams every five minutes, adds captions and logos, and queues them for posting to X, Instagram, and TikTok. The video is a live behind-the-scenes look at the technical architecture, honest failures, and the broader vision of a one-hour-a-day creator flywheel powered by AI agents.
Key points
1
Gaia Clipper uses a two-model chain (a 'picker' model and an 'editor brain') cycling through Claude, Gemini, Codex/GPT-5.6, and Grok to select and edit five-minute highlight clips from a live stream in real time.
2
Clean camera and screen ISO captures are achieved on an M1 Mac Studio using Apple ProRes hardware encoding, offloading transcode to an M3 Mac Studio Ultra via Thunderbolt 4 — a solution surfaced entirely by Claude via a slash-goal prompt.
3
Gemini's video understanding API (frame-by-frame analysis, up to 20GB upload or public YouTube URLs, 8 hours/day free) is used to analyze clips and generate an executive-level report on how to improve edits.
4
Mike ran an extensive AI thumbnail harness test (203 generations across GPT Image 2, Stable Diffusion/C Dream 4.5, Flux, Nano Banana, Gemini) and concluded C Dream 4.5 at 4 cents/generation rivals GPT Image 2 in quality.
5
A key honest finding: frontier models like Claude produce wildly inconsistent results day-to-day due to server load and undisclosed prompt/quant changes, making local models more appealing for repeatable automated workflows.
Actionable insights
→
Use a two-model architecture for AI video clipping: one model selects the highlight moment, a separate 'editor brain' model decides how to cut and style it — this separation improves output quality.
→
Gemini Flash video understanding can analyze YouTube videos via public URL for free (up to 8 hours/day), making it a powerful zero-cost tool for content analysis and chapter generation that most creators are not using.
→
Before committing to one AI image model for thumbnails, run a harness test across multiple models with identical prompts — C Dream 4.5 (Chinese model, 4 cents/generation) matched GPT Image 2 quality in Mike's 203-generation test.
→
Schedule heavy Claude Code tasks for early morning (US night-time) to exploit lower server load — Mike reports dramatically better performance before the US wakes up.
→
Use Assembly AI's universal transcription model to feed real-time transcript segments to a clipper agent, enabling time-accurate clip boundary detection and eliminating mid-word cuts.
Notable quotes

“The secret sauce in building anything right now with AI is how you stitch AI together, how you get it doing tasks. That is definitely the secret sauce and that is not an easy nut to crack.”

“One day AI is like phenomenal. The other day it's like split personality. And I don't understand it — it needs a lot of human taste to make it good.”

“First thing in the morning, flip open the laptop, play with Claude Code. It's lightning fast. It does everything. Amazing. It's brilliant. It's because my wonderful friends in the US are all sleeping.”

Worth watching?
✅
Worth watching the full video?
Watch if you want to see a creator actually building and debugging a real AI clipping pipeline live — the honest failures, architecture decisions, and model comparisons are more practically useful than most polished AI tutorials.
Topics
AI & TechClaude

Explore more summaries on these topics →

Saved you some time? The creator still deserves a like.

Watch on YouTube →
More like this
Claude Code Artifacts Are Here (No Backend!)
Creator Magic
Claude Code Artifacts Are Here (No Backend!)
Claude Code Artifacts — previously limited to Teams and Enterprise plans — has rolled out to all paid Anthropic subscribers, letting you generate shareable, interactive web pages directly from your coding session with no backend or deployment required. The creator demonstrates real use cases including repo explainers, Kanban boards, YouTube analytics dashboards, and AI thumbnail model comparisons.
2 min · 3 Jul 2026
Claude Tag Tutorial: AI Agent Living in Slack
Creator Magic
Claude Tag Tutorial: AI Agent Living in Slack
Claude Tag is Anthropic's new Slack-native AI agent that operates as a genuine team member with its own identity, credentials, and persistent memory — not just a chatbot. This tutorial walks through setup, channel-level customization, GitHub integration, and connecting Google Workspace tools so the agent can autonomously handle tasks like merging pull requests, sending emails, scheduling meetings, and recording transcripts.
3 min · 26 Jun 2026
Claude Fable 5 Runs My Entire Life (5 Builds)
Creator Magic
Claude Fable 5 Runs My Entire Life (5 Builds)
The creator demonstrates five real-world automation builds using Claude Opus 4 (referred to as 'Fable 5') across Claude Chat, Claude Cowork, and Claude Code — covering inbox cleanup, invoice automation, a DIY email marketing platform replacing a $149/month SaaS, AWS security auditing, and a live server monitoring dashboard. The video argues that the free access window (until June 22nd) makes now the ideal time to automate entire business operations.
3 min · 12 Jun 2026

Found this by searching? That's the hard way.

summree watches the AI channels you follow and sends you the summary the moment a video drops. Nothing gets past you. From £4/month.

Try it free