AI content machines fix slop by fixing human inputs first

Jul 20, 2026 · Lenny's Podcast
🎧 PodShort 47 min squeezed to 2 AI SprinklerAS Marketing Automation
Episode artwork
Alex Lieberman
Co-founder & Executive Chairman at 10x
Clara Bell
Product Leader & AI Obsessive at How I AI
Lenny's Podcast
47 min squeezed to 2
Full episode from Lenny's Podcast
Quotable Moments

AI slop is hilariously people just pointing the finger at themselves and saying I'm not intelligent enough.

Alex LiebermanCo-founder & Executive Chairman at 10x
01:13

The only time in my view that the content machine actually produces slop is more of an indictment of the person not sharing good enough ideas during the interview step, than the AI writing bad stuff.

Alex LiebermanCo-founder & Executive Chairman at 10x
01:02

The number one thing that's going to jump them out of this company is their inability to talk about all the amazing stuff that they're doing and build their own personal brands.

Clara BellProduct Leader & AI Obsessive at How I AI
05:59
Key Insights
  • AI is inherently bad at writing content "out of the box" and tends to produce "slop" unless carefully managed.
  • When AI-generated content is perceived as "slop," it is often more an indictment of the person providing the input or not giving good enough ideas during the interview step, rather than the AI itself writing poorly.
  • Creating content has provided many opportunities, but there's a cap on the amount of time that can be spent creating content daily, leading to the need for AI-native or AI-assisted content processes.
  • A major problem is how to make it as easy as humanly possible for employees to become content creators, even while they have full-time jobs, which the content machine aims to solve.
  • The initial step of the content machine, called 'The Oracle,' scans the last seven days of information from all channels a user is active in, ranks a shortlist of ideas based on a scoring system, and suggests content spikes.
  • By structuring content creation into a multi-step process and incorporating AI as a co-pilot, not just a driver, it's possible to raise the ceiling of content quality, especially for tasks outside of direct drafting.
  • A crucial step before implementing AI in any process is to first map out the entire existing workflow, assuming no constraints, to identify inefficiencies and areas for improvement, which often reveals significant waste even before AI is introduced.
  • The belief that "if you build it, they will come" is often not the case in a world where distribution is more important than ever, making proactive content creation and promotion essential.
Metrics Mentioned
  • 25% of time (Alex Lieberman can spend 25% of his time creating content every day.)
  • 15 content spikes (Alex Lieberman receives 15 content spikes daily from The Oracle.)
  • 7 days (The Oracle scans the last 7 days of information from all channels.)
  • 10,000 free credits (Firecrawl offers 10,000 free credits with code 'HOWIAI'.)
  • 1 million developers (Over 1 million developers use Firecrawl, including Clara Bell.)
  • 9,000 brands (More than 9,000 brands use Customer.io.)
  • 40% of inbound leads (The 'Own the Internet' campaign at StoryArb drove 40% of all inbound leads in one quarter.)
  • 5,000 dollars (Prize money offered in the 10x Creator Cup challenges.)
  • 70% of employees (A weekly challenge in the 10x Creator Cup requires at least 70% of employees to post once to unlock a prize.)

RevBots.ai View:

  • AI Sprinkler teams bolt on content AI without fixing input quality first.
  • ARM-stage orgs would integrate this workflow into their revenue stack.
  • Employee advocacy scales demand gen but requires guardrails.
  • Tab Hoppers lack the infrastructure to operationalize this approach.
🎧Full Episode:Lenny's Podcast →