Meeting Summary: AI Automation Strategy for Really American Media Operations

Overview

Justin Horwitz introduced Gary Sheng to Kenny Hesse, Really American's content producer, to discuss implementing AI solutions for streamlining content production processes. The meeting focused on identifying pain points in story selection, thumbnail creation, and content production workflows, with Gary proposing initial automation solutions including a thumbnail generator and story bookmarking system.

Key Topics Discussed

Introduction & Context Setting

• Gary's Background: Justin introduced Gary as "an operational genius" who graduated #1 from a selective AI software engineering bootcamp (100 hours/week)
• Previous Collaboration: Gary and Justin first connected through Facebook DM relationship when sharing high-performing memes
• Really American's Scale: Growing 100K+ YouTube subscribers monthly, generating $30-50K/month from YouTube, $20-30K from Facebook
• Vision for 2028: Building network model for "most insane news cycle ever" (2028 election)

Current Production Workflow Analysis

• Kenny's Process Overview:

  • Story selection from X (Twitter), Raw Story, Daily Beast, Instagram
  • Focus on underreported stories or unique angles
  • Scripting and bullet points for teleprompter
  • Title optimization using vidIQ plugin
  • Thumbnail creation in Photoshop
  • A/B testing (but not analyzing results)

• Time Requirements:

  • Minimum 20 minutes from download to upload
  • Typically 30-45 minutes for full process
  • Thumbnail creation is major bottleneck
  • Tony (thumbnail creator) must download video, find context in Slack, source images, create in Photoshop, get team approval

AI Automation Opportunities Identified

1. Story Selection & Discovery

• Sources to Monitor: X/Twitter accounts, Reddit, Raw Story backend data, subreddits
• Selection Criteria: Time sensitivity, underreported angles, "fucking around and finding out" stories
• Kenny's Insight: "If something's popping off currently or about to, and we're able to pick up on that and get a video out on it quick enough, then that has a lot more upside potential"
• Challenge: Current story aggregator lacks editorial "knack" - "he just puts a link and 'oh, you could talk about this'"

2. Thumbnail Generation

• Current Pain: Manual Photoshop process with multiple approval steps
• Justin's Vision: "Picture one, Trump angry. Picture two, Trump sad. Picture three, some lady jumping out with text"
• Proposed Solution: AI system where users specify elements and it generates thumbnails automatically
• Kenny's Note: "We should be able to simplify the thumbnails we make"

3. Script & Outline Creation

• Need: Quick bullet points from story selection to teleprompter-ready content
• Gary's Insight: Transform from "okay, we like your story" to "we have a great outline" very quickly

4. Data Analysis & Learning

• Current Gap: "We're just accruing data that no one's using"
• Kenny's Request: "I would love to have something run through our channel and gather data and statistics"
• Need: Analysis of top performing videos to identify patterns

X (Twitter) API Discussion

• Cost Structure:
  • $200/month for testing
  • Can scale to $15K/month for full usage
  • 15K post retrieval limit
• Justin's Perspective: "I could see it being worth a few grand a month if it's really providing a great [service]"
• Strategy: Monitor specific high-value accounts rather than broad scraping

Gary's Proposed Solutions

Custom Bookmarking Tool Demo

• Functionality: Save tweets/content with one click, add notes, create shared video idea repository
• Gary's Quote: "Your team having like a shared video idea app that connects to the same sort of production pipeline"

Initial Deliverables Plan

1. Thumbnail Generator: Based on analyzed patterns from top channels
2. Story Idea System: Integrated bookmarking and curation tool
3. Channel Analysis: Deep dive into performance patterns
4. Process Integration: Connect tools into unified workflow

Competitive Landscape & Urgency

• Competition Examples: Midas Touch, Keith Edwards (growing very fast)
• AI Threat: "How am I supposed to compete with George Clooney AI?"
• Time Pressure: Need to scale operations before 2028 election cycle
• Justin's Channels: Really American, US Democratic Socialists, Democratic Wins Media, Raw Story partnership

Key Quotes

Justin Horwitz

• "Gary's made like 20 different AI apps in his free time"
• "What I'm trying to figure out is, is there a way... to use AI technology to streamline our processes"
• "How can it make story selection easier, how can it make bringing up thumbnails [easier]"
• "If there's any way, especially now that we have some data... finding a way to aggregate and try to have AI maybe pick the best stories"

Kenny Hesse

• "I spend a lot of time figuring out what story I'm going to cover"
• "We're also A/B testing a lot between titles and thumbnails, but we rarely go back and look at the data"
• "From downloading a finished video to uploading it and publishing it... the quickest I can get it done is 20 minutes"
• "I would love to have something run through our channel and gather data and statistics"
• "We've never really gone through and like, here are our top performing videos of all time, here are what they all have in common"

Gary Sheng

• "The best outcome from this conversation would be to just understand what's annoyingly manual for you right now"
• "I wouldn't undervalue the human curation part"
• "The value is all in like integrating these little tools together for an actual business"
• "Before automating that, you can just manually spot it"

Action Items / Next Steps

Kenny's Deliverables

1. Provide list of Twitter accounts they follow for story sourcing
2. Share Tony's thumbnail breakdown guide sent to editors
3. Analyze 3-4 top performing channels (Midas Touch, Keith Edwards, etc.)
4. Document top performing videos with outlier scores
5. Send all documentation by end of day or next day

Gary's Commitments

1. Review provided materials and examples
2. Begin with thumbnail generator as first tool
3. Consider X API implementation ($200 test budget)
4. Join Slack channel for team observation
5. Get added to YouTube as analyst for data access

Justin's Actions

1. Add Gary to Really American Slack
2. Provide YouTube analyst access
3. Consider company credit card for API expenses
4. Facilitate ongoing collaboration

Strategic Insights

Operational Philosophy

• Speed Matters: "As soon as a new story pops, you gotta start making the video immediately"
• Data-Driven Decisions: Move from intuition to analytics-based content selection
• Human + AI: Maintain human editorial judgment while automating repetitive tasks
• Simplification First: "We don't need this three picture, somebody saying something every single time"

Success Metrics

• Reduce upload time from 30-45 minutes to under 20 minutes
• Automate story discovery across multiple platforms
• Generate thumbnails programmatically
• Analyze and apply learnings from A/B test data
• Scale content production for 2028 election cycle

Long-term Vision

• Create standardized production pipeline
• Enable any team member to produce content efficiently
• Build AI-powered producer for outline creation
• Develop data-driven content strategy
• Scale multi-brand, multi-platform operation seamlessly