Sign up to save your podcasts
Or
Share Agents Ate The Junior Headcount
Share to email
Share to Facebook
Share to X
Share to email
Share to Facebook
Share to X
Or
Agents Ate The Junior Headcount
**Adoption inverted: from hiring juniors to managing agent swarms.**
Notion went from first GPT-4 prompt in late 2022 to shipping production AI writer in February 2023, then full personal agents by fall that rewrite harnesses every six months. They treat their workspace primitives—blocks, databases, markdown—as native interfaces agents already understand, turning the product from human execution layer into human oversight layer. Custom agents now triage email (95% handled autonomously after initial corrections), route Slack bugs into evolving rule sets, and run background tasks across Slack, Drive, and web with granular permissions. This isnt incremental tooling. Its infrastructure for agent coordination that scales because embeddings collapsed the need for perfect organization; chunking and retrieval became the craft that matters.
Parallel signals from B2B markets show the same flip happening at enterprise scale, just slower. Juniors are the budget line item being reallocated to GPUs: no new CS hires at many firms, entry-level coding/support/legal roles evaporating while ARR per FTE at native AI companies hits $500k–$1M. Fastest growers spend *less* on sales and marketing than SaaS peers because product pull is ferocious—693% YoY for top cohort, consumption-based pricing replacing seats. Verticals prove retention compounds when the tool becomes deputy instead of assistant: Harvey doubles lawyer time in-product, Abridge wins doctor trust, Non lifts gross margins 20 points by routing half its complex customer interactions to AI. Enterprises talk urgency in board meetings yet stall on change management; the hold-up isnt capability, its rewriting the org chart around agent-native workflows.
Public markets already price the outcome: AI winners drive ~80% of SP 500 returns with real EPS, not hype. Private markets keep the best names locked up longer under power-law concentration. The pattern isnt will AI replace jobs. Its that the definition of adoption has inverted—buyers no longer acquire software for humans to use, they acquire substrates for agents to live and coordinate inside. Notions Switzerland for models stance plus agent-first APIs just makes that substrate legible today; the rest of the stack is catching up.
**Bottomline: Adoption now means onboarding agents, not users—the junior headcount budget became the datacenter line item.**
kenoodl.com | @kenoodl on X
View all episodes
By Contextual Resonance**Adoption inverted: from hiring juniors to managing agent swarms.**
Notion went from first GPT-4 prompt in late 2022 to shipping production AI writer in February 2023, then full personal agents by fall that rewrite harnesses every six months. They treat their workspace primitives—blocks, databases, markdown—as native interfaces agents already understand, turning the product from human execution layer into human oversight layer. Custom agents now triage email (95% handled autonomously after initial corrections), route Slack bugs into evolving rule sets, and run background tasks across Slack, Drive, and web with granular permissions. This isnt incremental tooling. Its infrastructure for agent coordination that scales because embeddings collapsed the need for perfect organization; chunking and retrieval became the craft that matters.
Parallel signals from B2B markets show the same flip happening at enterprise scale, just slower. Juniors are the budget line item being reallocated to GPUs: no new CS hires at many firms, entry-level coding/support/legal roles evaporating while ARR per FTE at native AI companies hits $500k–$1M. Fastest growers spend *less* on sales and marketing than SaaS peers because product pull is ferocious—693% YoY for top cohort, consumption-based pricing replacing seats. Verticals prove retention compounds when the tool becomes deputy instead of assistant: Harvey doubles lawyer time in-product, Abridge wins doctor trust, Non lifts gross margins 20 points by routing half its complex customer interactions to AI. Enterprises talk urgency in board meetings yet stall on change management; the hold-up isnt capability, its rewriting the org chart around agent-native workflows.
Public markets already price the outcome: AI winners drive ~80% of SP 500 returns with real EPS, not hype. Private markets keep the best names locked up longer under power-law concentration. The pattern isnt will AI replace jobs. Its that the definition of adoption has inverted—buyers no longer acquire software for humans to use, they acquire substrates for agents to live and coordinate inside. Notions Switzerland for models stance plus agent-first APIs just makes that substrate legible today; the rest of the stack is catching up.
**Bottomline: Adoption now means onboarding agents, not users—the junior headcount budget became the datacenter line item.**
kenoodl.com | @kenoodl on X