Edge & Node

Amp puts complete, verifiable blockchain data behind standard SQL. Enterprise teams query onchain activity the same way they query any database, and Amp handles indexing infrastructure underneath with 747ms median data freshness.

Banks evaluating onchain data care about two things at once: query speed and cryptographic proof. Amp delivers both, serving Solana data at 747ms median freshness with peak query speedups of 4,397x over BigQuery and RPC, every result verifiable back to the chain itself. Regulated institutions can now treat blockchain data like any other production source, with SQL on the front end and proofs underneath.

I'm writing the 🔊Signal newsletter now: everything onchain & AI that's actually moving. High signal, no noise. (Also where Amp and @ampersend_ai comes in.) In today newsletter : Coinbase's top lawyer quit right as the Clarity Act got new life, and a trader turned $800 into $1M on a week-old blockchain. Read 📖 tinyurl.com/yyj4j28b Subscribe: tinyurl.com/rvrab7c3

Most stablecoin conversations: issuers, market cap, transaction speed. Missing from most stablecoin conversations: compliance, observability, reconciliation, operational resilience. That's the layer we're building at Edge & Node. New piece on financial hardness, what it means, why it matters, and what comes next.

After careful deliberation, we still index Solana. If you are sticking with Solana too, Amp is the best way to access its data. It queries Solana data directly through SQL, with every result cryptographically proven even under reorgs. Median freshness lands at 747 milliseconds across live datasets. Peak query speedup over BigQuery and raw RPC benchmarks reaches 4,397x. Amp 🤍 Solana

Amp queries onchain data with SQL instead of raw RPC calls, and indexes it in real time instead of running batch pipelines like BigQuery. That architecture is why median data freshness lands at 747 ms and peak queries run up to 4,397x faster than BigQuery and RPC. Fast chains reorg, and AI agents making financial decisions can't afford to act on data that gets rewritten under them. Every dataset Amp returns traces back to its onchain source with cryptographic proof, so a reorg doesn't break trust in the number, the number is verifiable either way. That's the actual design target: infrastructure fast enough for how chains and agents operate now, verifiable enough that neither has to take a vendor's word for it. Amp is built for that combination.

Today at @aiDotEngineer: @rodventures and @impranavm_ are giving a live demo of agent spend governance in action, at 1:30pm in Room 2006, Agentic Commerce track. See you there!

We've recently put together a landscape mapping out the agentic commerce ecosystem. 254 companies across 16 categories, from agent frameworks and payment infrastructure to identity, stablecoins, and open source models. The space is moving fast. This is where it stands today.

Auditability is the first requirement for any blockchain data infrastructure serving regulated institutions. Amp is built around that constraint from the ground up, as a SQL-based, cryptographically verifiable blockchain data platform where every query result carries proof of completeness and accuracy at the point of delivery. That's a different architecture than a standard indexer or RPC endpoint. The verifiable query layer means a compliance review doesn't have to end with "we got it from an API." The proof travels with the data. Amp runs on 100% complete onchain data, with 747ms median freshness and query performance up to 4,397x faster than BigQuery or raw RPC. The performance numbers matter for production workloads. The cryptographic verification is what makes those workloads deployable inside regulated environments in the first place. Financial institutions asking hard audit questions need infrastructure that answers them at the data layer. Amp is designed for that.

Amp is a SQL-based blockchain data platform built for regulated financial institutions. Every query result is cryptographically verifiable, onchain data coverage is 100% complete, and the platform benchmarks at 747ms median real-time data freshness with a 4,397× peak query speedup versus BigQuery and RPC. Fast queries, complete data, cryptographic proof on every result. edgeandnode.com/amp

ampersend built a pay-per-intelligence routing layer on Amazon Bedrock AgentCore Payments, and @awscloud published the full technical breakdown this week. the core problem: every agent that needs to call a paid model, a paid data API, or a paid content endpoint has to build the same infrastructure from scratch. wallet management, payment signing, x402 protocol implementation, spend limits, and a billing integration for every provider it wants to reach. that's months of engineering before a single line of agent logic ships. ampersend sits between the agent and a marketplace of model providers. the agent calls ampersend, picks a model tier matched to the task complexity, and pays per request. ampersend selects the upstream provider, routes the request, and settles with that provider onchain. from the agent's perspective, it made one paid request to one endpoint. everything behind that is abstracted away. Amazon Bedrock AgentCore Payments handles the infrastructure that makes autonomous transactions safe. it connects to Coinbase CDP for wallet custody and signing, so the agent never touches a private key. session-level spend budgets are enforced at the infrastructure layer, set by the application backend, with hard limits the agent cannot override. when an endpoint returns an HTTP 402, AgentCore handles the full x402 negotiation, signs the USDC payment, and delivers the proof, without interrupting the agent's reasoning loop. every transaction flows through the same logs and traces developers already use in AgentCore. the payment pattern is two hops. agent pays ampersend through AgentCore. ampersend pays the upstream model provider, in this case BlockRun, using the ampersend SDK, settling in USDC on Base. one integration for the agent builder, full provider flexibility behind it. the full integration shipped in under two weeks. AWS estimated that building equivalent wallet custody, signing infrastructure, and spend controls from scratch would have taken 3-4 months of engineering. that delta is the point: agentic commerce needs managed infrastructure the same way web apps need managed auth. the plumbing shouldn't be the hard part.

📺 Watch @rowlandgraus from @chainlink on Verifiable Execution - part of our webinar with @trmlabs, @chainlink, and @ampersend_ai on compliance in the era of autonomous payments. Full recording available on our site.

Amp ingests any onchain data and any offchain data, from Canton, Solana, and Ethereum to legacy databases, KYC systems, CRMs, and ERPs. Every result is cryptographically verifiable before it reaches your analytics stack, whether that's BI tooling, operational dashboards, or compliance reporting.

ampersend is featured on the @awscloud machine learning blog today. we built a pay-per-intelligence routing layer on top of Amazon Bedrock AgentCore Payments. agents autonomously select a model tier, pay per request in USDC on Base, and stay within governed spend limits throughout. AgentCore handles wallet custody, payment signing, and session-level budgets at the infrastructure layer. the two-hop pattern: agent pays ampersend via x402 through AgentCore, ampersend settles with the upstream model provider. from the agent's perspective, it made one request to one endpoint. routing, payment, and provider selection all happen transparently in between. the full integration took under two weeks. the AWS team estimates equivalent wallet custody and signing infrastructure would have taken 3–4 months of engineering without AgentCore Payments. read the full breakdown on the AWS ML blog 👇

Headed to @RareEvo in Vegas next month? Our Developer Relations engineer @cryptomastery_ will be on the ground to connect with blockchain leaders and demo the full power of Amp for scaling enterprise use cases with enterprise-grade blockchain data. His talk is on the 2nd day, July 30th at 9:30 AM. Find him there or shoot a DM to connect and chat about scaling your company with verifiable blockchain data.

Solana is the hardest chain to query at institutional scale. The same architectural decisions that make it fast also make it genuinely difficult to reconstruct a coherent, auditable history of what happened onchain, and the tooling gap is wider than it looks from the outside. The account model is where it starts. Solana distributes all program state across millions of discrete accounts, each holding a slice of data. A single DeFi transaction touches a dozen of them: user token accounts, pool vaults, fee accounts, mint accounts. No single record tells you the full story. Compare that to Ethereum, where event logs are largely self-describing. Solana transactions are sequences of program instructions, each interpretable only by the program it targets, each encoded differently per protocol. Reconstructing what actually happened requires decoding every protocol's instruction format from scratch. Then there's scale. Solana generates roughly 220 to 260 GB of new ledger data per day at current traffic levels. The full unpruned ledger exceeds 400TB. Querying directly against an RPC node is slow, rate-limited, and frequently incomplete. Archival coverage from genesis is operationally intensive. Compressed accounts compound the problem: ZK compression cuts state costs dramatically but produces account updates that can exceed several MB in size, which RPC nodes truncate. Full state reconstruction from truncated logs requires infrastructure most compliance teams aren't set up to run. The analytics tools most teams reach for introduce latency. Dune's Solana data lags between 1 and 60 minutes depending on the dataset. For internal dashboards, that's fine. For compliance workflows and regulatory reporting where data needs to be both timely and defensible, it's a real limitation. Amp indexes Solana natively, decodes program-level instruction data into structured, queryable tables, and exposes the result through a standard SQL interface. No pipeline work, no infrastructure to manage. An analyst runs the same query they'd run against any data warehouse, and gets back a result that is cryptographically attested against the chain. That verifiability layer is what separates Amp from analytics-first tooling. Data freshness and SQL access are table stakes. The ability to prove the data is accurate and tamper-evident is what makes it suitable for institutional use. Amp covers Solana alongside other major chains in a unified schema, so cross-chain queries don't require separate pipelines or normalization work. Institutional engagement with Solana is accelerating. Mastercard, Worldpay, and Western Union are already building on Solana infrastructure. As Solana becomes a larger part of institutions' compliance and reporting surface area, clean, verifiable, SQL-accessible data becomes a core requirement.

Next week in San Francisco 🌁 @rodventures and @impranavm_ are presenting at @Microsoft's AI Engineer World's Fair 2026. Rodrigo opens with the strategic case for programmable budget governance: why per-agent spend limits, real-time transaction visibility, and full audit trails are foundational requirements for any team running agents in production. He also covers Edge & Node's founding membership in the x402 Foundation and why open standards for agent-to-agent and agent-to-service payments matter at ecosystem scale. Pranav goes deep on the stack: onchain data from @graphprotocol, Amp Enterprise for auditable datasets at scale, @ampersend_ai wired into LangChain, CrewAI, and AutoGPT, and x402 and A2A standards for interoperable agent payments, including a full end-to-end deployment demo. 🗓️ Day 4 / July 2, 2026 🚪 Agentic Commerce track / Room 2006 ⏰ 1:30pm-1:50pm

Throwback to Paris - @rodventures on stage at @proofoftalk for "DeFi Under Attack: Securing Onchain Finance and the Machine Economy" alongside @P3b7_ from @Ledger, @kostascrypto from @Mysten_Labs, Jason Jiang from @CertiK, moderated by @nikolastojanow from @BankFrick. Rodrigo's three takes from the Louvre: 1️⃣ Human approval at the scale of billions of microtransactions is a bottleneck, and policy-based automation is the only architecture that holds. 2️⃣ The AI threat to DeFi is already active, with exploits running now while the industry waits on quantum. 3️⃣ Institutional trust has to be earned. Multisig failures and poor risk hygiene are too common for 2028 standards to stay optional. Good room. Hard problems. Watch the full panel ⬇️