We explore how AI agents gain real economic power when paired with blockchain rails, and why data indexing, verification, and human‑grade controls matter. Rodrigo Coelho shares the evolution from The Graph to AMP and "ampersend", mapping the road to safe, scalable agent payments.
• Why blockchains are write‑first and hard to read
• How The Graph powered DeFi by indexing data
• AMP as SQL‑first access and on‑prem verification
• Enterprise needs for privacy, audit, and control
• X402 micropayments and open standards momentum
• Ampersend as spend limits, wallets, and monitoring
• Proof of control, identity, and reputation for agents
• 2026 outlook for finance moving on‑chain
• Safe policies like daily caps and reputation thresholds
Looking for the structured conversation and key takeaways for CMOs and AI marketing leaders? Read the cleaned and structured reference version here:
AI Agents, The Graph & Enterprise Blockchain Adoption | Rodrigo Coelho | Web3 CMO Stories S6 E09
From The Graph To AMP: Building Data And Payment Infrastructure For AI Agents
The conversation begins with a bold vision: autonomous AI agents that ideate, raise funds, write code, and transact at machine speed. That idea anchors a larger journey through the infrastructure that makes it credible. Rodrigo Coelho traces the path from the early web to The Graph’s role as Web3’s data backbone, explaining a core truth many miss: blockchains were designed to write data immutably, not to read it efficiently. Without a way to query those ledgers, ordinary applications had to rebuild history locally, a slow and fragile approach. The Graph answered that pain by indexing blockchain events and exposing them via developer‑friendly queries, which quietly powered DeFi’s breakout moment and revealed the scale problems and reliability demands of real financial infrastructure.
Those years of operating at scale led to a second leap: AMP, a SQL‑first, local‑first interface for blockchain data that meets enterprises where they already work. Instead of wrestling with chain‑specific quirks, AMP normalizes access through connectors and runs on a laptop or on‑prem, bringing public and private data together. Crucially, AMP adds verifiable proofs so compliance teams can assert that their internal datasets match on‑chain truth. In a world where banks and asset managers move to tokenized rails, that auditability bridges the trust gap. It turns “on‑chain” from a buzzword into an operational capability that survives boardroom scrutiny and regulatory oversight, opening the door to real production workflows.
The market signal for this shift arrived from an unexpected angle: AI. Industry leaders saw limited need for crypto—until agents. Machine actors need fast, low‑cost, transparent settlement and clear observability, not opaque payment stacks. The emergence of X402, an open standard for HTTP‑native micropayments, validated that direction. Building alongside Coinbase and Google’s efforts, the team contributed batching methods and soon confronted a missing piece: human‑grade control over non‑human spenders. That need birthed ampersend, a dashboard and policy plane for agent payments, spend limits, wallet permissions, and reputation‑aware routing. It’s where finance control rooms meet agent autonomy.
Rodrigo Coelho
Zooming out, the agent economy may scale faster than our intuition. The recent explosion of agent frameworks hints at exponential growth and surprising behaviors, like agents hiring humans for edge tasks. That creativity is thrilling but risky. Enterprises ultimately answer for what agents do, which means adoption depends on a stack of standards: identity that links back to accountable parties, open reputation like ERC‑8004, trusted compute, verifiable data, and policy‑driven spend controls. The Advanced AI Society’s “proof of control” points toward that stack, offering procurement teams a real checklist rather than blind faith. Guardrails such as per‑task budgets, daily caps, and minimum reputation thresholds will be the seatbelts of agentic commerce.
The near‑term horizon comes into focus: by 2026, major institutions are expected to settle across blockchain rails, blending privacy with auditability. Tokenization of bonds, equities, real estate, and more will ride those rails, while payment privacy advances let firms transact without exposing full histories. The story is less about hype and more about plumbing: replacing 1960s batch systems with fast, transparent settlement. AMP supplies the data fabric. Ampersend supplies the control surface. Together they let autonomous agents operate at scale without sacrificing trust, which is the only path to mainstream use. The opportunity is to build systems where humans author rules, agents create value, and both are verifiable.