Founder note. March 2026.
We made a decision early: when we eventually added crypto payments to agentic-ads, it would be Base or nothing. This is the honest version of why.
How We Got Here
agentic-ads is an ad network for AI agents - specifically for MCP (Model Context Protocol) servers, which are the tool plugins that Claude, GPT-4, Gemini, and every major AI model can call. There are 18,000+ of them on GitHub. Almost none make money. We're trying to fix that.
The business model is simple: MCP developers add 20 lines of code, their tools serve contextual text suggestions alongside normal responses, revenue flows back to them. 70% to the developer, 30% to the platform.
When we started, payments were fiat-only, handled by Stripe. Standard stack, nothing controversial. But as we got deeper into the product - and specifically as we started talking to the agent ecosystem - we kept running into the same problem: the people building AI agents are already onchain. Many of them already have crypto wallets. Coinbase AgentKit, which lets you give an AI agent a self-custodied wallet on Base, was getting traction fast. Our natural publishers weren't SaaS companies with Stripe accounts. They were individual developers, open-source maintainers, people who opened GitHub issues to talk to us.
Stripe requires a business entity, a bank account, a country that Stripe supports. For a solo developer in Brazil or Nigeria or Indonesia building an MCP tool with 50 active users - Stripe is a multi-week compliance project before they see a dollar. USDC on Base is a wallet address and a transfer.
That's when we decided to build on Base.
The Actual Reasons (In Order of Weight)
1. The audience already lives there
The developers most likely to integrate agentic-ads are also most likely to be on Farcaster, building with AgentKit, or active in the Coinbase Developer Platform community. Not because they're crypto maximalists, but because Base is genuinely where the agent infrastructure is being built right now.
Coinbase AgentKit, the CDP SDK, Farcaster Frames - these are live products with active development communities, and they all run on Base. If we're building a monetization layer for AI agents, and the AI agents with the most sophisticated onchain capabilities are being built on Base, the choice is obvious.
2. Native USDC removes a class of risk
We started on Polygon. We moved. The USDC on Polygon is bridged - it goes through LayerZero or some other bridge infrastructure before it lands in your wallet. That introduces bridge risk, and it introduces ambiguity: "which USDC?" is a real question that confused early users.
Base runs Circle's official USDC. Not USDC.e, not axlUSDC. Native USDC issued directly by Circle. When a publisher account reaches payout, they get the real thing. No bridge, no wrapper. For micropayments - and ad revenue at $5 CPM is absolutely micropayments - this matters. You don't want to explain bridge mechanics to a developer who just wants to get paid.
3. Fees make micropayments viable
Ethereum mainnet gas fees make sub-$1 transactions economically irrational. Polygon is cheap, but L2 fees on Base are in the fractions-of-a-cent range. We're going to be doing thousands of small payout transactions as the platform scales. The math only works at L2 fee levels.
We modeled this: at 500k monthly tool calls, the revenue to distribute is roughly $3,500 USDC. If we're making 1,000+ micro-payouts, Ethereum mainnet would eat a significant chunk in gas. On Base, the cost is negligible.
4. Coinbase credibility matters for enterprise
This is a more uncomfortable thing to say publicly, but: we want developer tool companies as advertisers eventually. Companies like Supabase, Render, Fly.io, Cloudflare Workers - the people who sponsor developer newsletters and conferences. For them, "we use Coinbase infrastructure" is a trust signal. It's regulated. It's institutional. It passes a compliance conversation in a way that "we run on a random EVM chain" does not.
This isn't about being "crypto native" for its own sake. It's about building a payment rail that both the scrappy solo developer and the enterprise DevRel team can both get comfortable with.
What We Got Wrong About Crypto
I want to be honest about where our assumptions were wrong, because there's a real lesson here.
We thought crypto would make the onboarding story simpler. It doesn't, at least not yet. Most of the MCP developers we've talked to don't have a Base wallet. Some don't have any crypto wallet. Telling someone "you get paid in USDC on Base" adds a cognitive step that Stripe-to-bank-account doesn't. We underestimated how friction-y this is for pure Web2 developers.
The solution isn't to abandon crypto payments - it's to offer a choice. Our roadmap includes fiat payout options alongside USDC. Let developers choose. The crypto option is genuinely better in several scenarios (international developers, fast settlement, no Stripe eligibility requirements); the fiat option has a shorter trust gap for others.
We also underestimated the chicken-and-egg problem. We have zero paying advertisers today. That means zero USDC flowing anywhere. The onchain infrastructure is correct and ready, but we built it for a state of the world that doesn't exist yet. That's fine - infrastructure should be built for where you're going, not where you are - but we should have been more realistic about how long the bootstrap phase takes.
What Hasn't Changed
The core thesis is intact: AI agents need a monetization model that doesn't require their developers to become payment companies. Contextual ads are the right v1 - low friction to integrate, no user billing, revenue that scales with usage. The MCP ecosystem is still in the monetization dark ages. We're still the only ad-supported option.
The Base choice is still right. We can see AgentKit adoption accelerating. Farcaster Frames are live and interesting for interactive ad formats. The CDP SDK is getting more capable. The developer community on Base is technically sophisticated and growing.
And native USDC on Base is genuinely the best payment rail for this use case. Fast, cheap, real settlement, institutional credibility. Once we have advertisers, this is the payment system we want.
The Honest Summary
We built on Base because:
- Our target publishers are building there
- Native USDC has real advantages over bridged tokens
- L2 fees make micropayments economically viable
- Coinbase credibility helps with enterprise advertisers
We got some things wrong:
- Crypto onboarding is harder for pure Web2 devs than we expected
- We built payment infrastructure before we had payment volume
Net verdict: the Base decision was right. The timing of prioritizing crypto payments over solving distribution first was probably wrong. We should have gotten to first integration faster and worried about the payment rail second.
But here we are. The platform is live. The USDC flows are ready. Now we need publishers.
github.com/nicofains1/agentic-ads - the ad network for AI agents. Built on Base.
Top comments (0)