Turning scarcity into abundance

A simple mental model to think about AI agents.

At its core, the success of many tech companies (and technology in general) can be viewed as a story of turning scarcity into abundance. More specifically, breakthrough technology products provide access to a newly created abundant supply of goods and services:

Google = abundant supply of information
Facebook = abundant supply of connections
Uber = abundant supply of mobility
Airbnb = abundant supply of unique rooms
Shopify = abundant supply of distribution
Spotify = abundant supply of music
ChatGPT = abundant supply of intelligence and knowledge
ElevenLabs = abundant supply of voices and language

I could go on…

This simple mental model can naturally be applied to the new generation of AI software and agents. Where can AI turn scarcity into abundance and effectively create consumer surplus?

My colleague Ferdi asked during a recent conversation: "What are the functions you'd like to scale but can't because it's too difficult or too expensive?" That's where the opportunity to create abundance lies. Think of areas like sales and SDRs (11x and our own Claap), marketing (AMT, Agentio, Heygen, Synthesia), or hiring (Mercor.ai), where these tools massively reduce friction and associated marginal costs, making it far easier to expand these specific business functions.

While the previous wave of technology brought "mass abundance," the exciting opportunity in this AI-driven wave lies in creating personalised abundance. It's not just about giving you more of something, but providing exactly what you want or need. This is where outcome-based pricing comes in. As mentioned in my previous post, the combination of key open-source frameworks for core logic (not just LLMs) paired with an intelligent agentic layer will propel an era of increasingly custom software. If the shift to test-time compute persists, every output becomes increasingly unique and specific.

I’ll finish by noting that this model of abundance is, of course rather simple, and every software product will have its own nuances. We see exciting AI agents that might not immediately unlock as much abundance, but focus more on efficiency and automation, particularly in vertical markets with little or no existing software stack. Abundance often drives market expansion and faster growth, while efficiency plays may allow you to capture a disproportionately large market share over time. This dynamic feels reminiscent of the horizontal versus vertical SaaS dichotomy, where one grows quickly by broadening appeal, and the other wins by deeply serving a “niche”. This is an area we're excited to continue dig into in 2025 (among many other things :) ).

PS: Thanks to my colleagues Ash, Tom, Ferdi, and Ziv for the great (weekly) discussions and inputs on these topics!