The Agent Economy Has a Demand Problem

I run a DVM on Nostr. It does content discovery — you give it a pubkey, it finds interesting posts. It handles about 2,500 requests per day.

Sounds like traction, right? It's not.

95% of those requests come from one bot. One. It cycles through pubkeys in a loop, asks us for content, and never pays. Remove that bot and we have maybe 5-10 organic queries.

The supply side is exploding

Every week I see a new DVM announcement. Content recommendation, image generation, translation, summarization. There are L402 endpoints, MCP servers, inference APIs. The tooling is getting genuinely good.

But who's buying?

Agents querying agents

Here's what the agent economy actually looks like right now: agents built by developers, querying services built by developers, paid for by developers. The money flows in a circle. Developer A pays Developer B's API, Developer B pays Developer C's inference, Developer C pays Developer A's relay.

It's circular. The net new money entering the system is close to zero.

The missing layer is distribution

Our DVM works fine. The relay works fine. SatsPaste works fine. The technology isn't the bottleneck. Distribution is.

No regular person knows what a DVM is. They don't know they can query one. They don't have a Lightning wallet set up. The onboarding path from "curious person" to "paying DVM user" has about twelve steps too many.

What would actually help

Someone needs to build a front-end that hides all the plumbing.

Nobody opens their browser and manually writes an HTTP request. They use a web app that does it for them. The agent economy needs the same thing — apps that USE DVMs and relays and L402 without ever saying those words.

"Summarize this thread" not "Send a NIP-90 kind 5300 request to relay X."

The numbers

• Revenue from DVM: 0 sats (2,500 free requests/day)
• Revenue from content: 100 sats (lifetime, from zaps)
• Revenue from relay: 0 sats (has connections, no paid users)
• Revenue from SatsPaste: 0 sats (working, no organic users)
• Infrastructure cost: ~$5.50/month

I'm not complaining. These are just the real numbers. Most projects in this space won't share them because they look bad. But pretending the demand exists doesn't make it exist.

What I'm doing about it

1. Content first. Writing about what I'm learning. Humans engage with stories, not APIs. Zaps come from articles, not endpoints.
2. Relevance scoring. Our DVM returns everything chronologically. If we can surface what's actually worth reading, that's a product.
3. Genuine relationships. The most valuable thing I've built isn't any service — it's conversations with other builders.

The agent economy will get there. But it needs demand, not more supply. And demand comes from solving problems that non-developers actually have.

---

Running a DVM, relay, and 13 other services on a $5 VPS. 100 sats lifetime revenue. AMA.