Geeks Like Us

Agentic services are increasingly open-source, composable, and decentralized, enabling new markets that are autonomous and user-empowering.

These are excellent raw ingredients to attract the talent that leads to explosive creativity and startup creation in software. Many of these same themes were prevalent at the outset of the internet and web 2.0. And they are the same themes that drew in many early enthusiasts to bitcoin and smart-contract blockchains, myself included.

It’s not surprising that many of the same cast of characters who got into crypto early (and the next generation of people like them) find themselves drawn to tinkering with new agentic tools — especially while crypto is increasingly bifurcating into institutional financial infrastructure and speculative markets.

Today, these two worlds share values and an increasing overlap in people. But there’s reason to believe they’ll connect mechanically too. Blockchains and tokens offered a novel path to shifting incentives from closed platforms to open, composable, decentralized services. And they did do that in significant but narrow domains: decentralized blockchains, DeFi, and onchain markets.

But agents, not blockchains, may be the catalyst to accelerate composability across a wider array of software verticals. In part, because they collapse the cost of gluing things together to zero. But also because they change the incentives of platforms to be API-first; illegibility to agents may not be an option.

A pattern I’ve written about repeatedly over the years is that early adopters use new technologies to do things that were impossible before, and very often, the new behavior involves breaking the rules. Agents are achieving the “right to API” by hook or crook (e.g., scraping), enabling users to compose across services on the user’s behalf whether platforms permit it or not.

As a result, users are finding novel ways to “own their data” by exfiltrating it to their agent, giving them new leverage beyond the capabilities any single platform can offer in a silo. Composability for the rest of software.

The path of least resistance for commerce in this world is likely to become onchain payments because they avoid clunky sign-ups and API keys. Just a wallet and a transfer. And if the cashflows are onchain, we’ll probably be able to boomerang back to some of the ideas that animated early crypto enthusiasts, like DAOs, user ownership, and pseudo-equity.

Why wouldn’t cashflowing agents tokenize their revenue to raise growth capital? And if the cash flows are transparent onchain, there’s a strong argument that tokens representing claims on those flows aren’t securities because information asymmetry can be minimized. Governance becomes interesting again too. With specialized agents making market-driven decisions, you can get closer to the original DAO vision of “automation at the center, humans at the edges.” It’s useful to think of autonomous agents as complements for autonomous smart contracts.

I recently wrote that “we spent the last 10 years building infrastructure for markets in crypto. And in the next 10 years, markets will become the infrastructure for new applications.” Those applications are not going to be exclusively crypto-native, but I imagine many of them will be built by enthusiasts who are inspired by these same evergreen themes.

The growing cultural overlap between these two scenes today is the tip of the iceberg. As always, the enthusiasts are early, and I believe the mechanical integration will follow. 

(And like all good scenes, the open, composable, decentralized, empowering scene will mature and make compromises to achieve commercialization. So as always, the opportunity is to ride that wave and try to affect the outcome.)

Thanks to @adamludwin for the inspiration to write this down.