The creator using AI to go viral | Will Weinbach

Summary

Will Weinbach describes how he went from using ChatGPT casually to building agentic systems that automate everyday work, especially social content production and distribution. A centerpiece is his autonomous social agent, which pulls content, edits it, writes overlays and captions, posts across platforms, and then uses analytics to improve itself. He and Parth Patil compare agent frameworks like OpenClaw and Hermes, emphasizing visibility, memory, and personalization as the features that matter most. The conversation also extends to consumer hardware experiments, like running an agent on a TV, and to enterprise use cases in Slack and group chats. Reid Hoffman argues that resumes may become more modular as AI certification and workflow validation become part of how capability is measured.

Notes

• Parth frames AI-native creator funding as $1,000 per week in tokens plus access to custom agents.
• Will says he still does not code, but now uses prompting to build automations and workflows.
• He treats ignorance of backend details as a creative advantage.
• Parth says OpenClaw went viral in February after launching earlier in the year.
• Will says OpenClaw lacked visibility, while Hermes gave him clearer insight into agent behavior.
• Hermes adds observable skill creation, improvement, and memory across long-running tasks.
• Will sees a future of specialized agents for different life and business functions.
• Parth describes Hermes as repeatedly updating skills and memory as it learns.
• Will's favorite project is an autonomous social agent with a full content pipeline.
• A prompt mistake made the company account post four times at 4 a.m.
• The workflow starts with a licensed content library and strict parameters on what the agent can pull.
• The agent has copywriting and overlay-generation skills tuned by top-performing examples.
• Ordinal acts as the autonomous content calendar and API layer for posting.
• Ordinal supports cross-posting across LinkedIn, TikTok, YouTube, and Meta/Instagram.
• Will says the most valuable distribution signals are average watch time and skip rate.
• The system closes the loop by revising content based on performance data.
• Will experimented with putting a Hermes agent on a television instead of buying a Mac Mini.
• He generalizes that AI could run on anything with Wi-Fi or Bluetooth.
• Parth says putting these systems in Slack or group chat lets multiple people shape the agent.
• Will argues older marketers may miss the broad, attention-economy logic of short-form social.
• His content framework centers on identity, connection, status, and emotion.
• He designs for the first three seconds and the payoff by the end of a nine-second clip.
• Will says his own attention span has shrunk and will likely keep shrinking.
• He thinks job screening may shift from resumes toward AI-tool capability.
• Reid says resumes may stay, but become more modular and backed by AI certification agents.
• Parth says there is still no textbook or class for AI workflow mastery, so hands-on experience is the edge.