AI Agents, 12,000 TPS, and Quantum Readiness: Insights from Sei Development Foundation's Justin Barlow
As blockchains mature, the conversation is shifting from "can it run?" to "can it automate at scale?" At Blockchain Futurist Conference 2025, Senior Editor Dr. Guneet Kaur sat down with Justin Barlow, Executive Director at the Sei Development Foundation, to dig into speed, agentic automation, and long-term security as quantum computing inches closer.
Barlow's core message: performance and developer experience matter right now, while the stack quietly gears up for an agent-first future and eventual cryptography upgrades.
12,000 TPS and the developer reality
Sei's pitch to builders is simple: a fast chain that feels like using a mainstream app. "When you make a transaction using the Sei network, it's fast and it's cheap," Barlow said. "For the user, it's a very smooth experience that feels just as easy as using something like Venmo, but has the benefits of blockchain on the back end."
On throughput, he was direct: "Sei is an extremely scalable, high-throughput chain. You can see upwards of 12,000 transactions per second without having the performance bottlenecks that you have on other chains." For engineers, that means lower UX friction, less queue anxiety, and fewer spikes derailing product flows.
Agents on-chain: MCP integration and early X402 support
Sei integrated MCP (Model Context Protocol) in July 2025 so AI agents can read chain data and execute transactions. Barlow expects a significant shift over the next year: "I think it's likely that not just on Sei but on the internet in general, we'll see a significant portion of transactions being done by AI agents."
He also pointed to Sei's early adoption of the X402 protocol for seamless agentic payments. "Sei, like a number of other technologically advanced chains, is making a really big effort to adopt protocols that can help enhance the experience for AI agents. An example is the X402 protocol, which was adopted by Sei very early on and has since been adopted by a number of other chains as well."
X402 tokens: signal vs. noise
The flood of "X402 tokens" after Coinbase's announcement didn't impress him. "The vast majority of these tokens have no underlying connection to X402 other than in name," he said. Utility comes later: "During the memecoin craze, tokens like SPX 6900 served as memes of the SPX in traditional finance, but now we are seeing the actual launch of tokenized equities on blockchain rails. Similarly, with X402, we have the memes first and later we'll have the real integrations."
Quantum: prepare without distraction
On quantum risk, Barlow urged a practical stance. "Many people are talking about quantum resistance, but what's important to remember is how far off we are from these things actually being used in production." Translation: ship today, stay ready to migrate tomorrow. "Teams like Sei Labs and others in the ecosystem can build a lot in the time between now and when quantum becomes a real threat."
If you're tracking standards, the NIST PQC process is the reference point. Expect staged upgrades, not an overnight switch.
Agentic rails for finance and apps
Barlow sees agent-first flows moving from novelty to default. "I think a lot of on-chain finance and real-world finance will move to agentic rails. And so what that means is that agents will be paying for everyday things like API calls, managing your data, etc."
Infrastructure follows the use case: "Sei is really gearing towards this place where you can have a fast, cheap experience, not just for real-world users but also for agents as well." He added that tokenized institutional funds (via KAIO on Sei) should be compatible with agent flows by design.
Trust, transparency, and operational safety
Sei leans on openness and proven throughput. "Sei is an open source, fully permissionless blockchain. And what that means is that anyone can verify transactions on the chain," Barlow said. The network has processed over 4 billion transactions and continues to attract teams that need speed and auditability.
On risk with AI agents-bugs, misbehavior, and hallucinations-Barlow stressed developer control. "There are risks across the spectrum when you talk about crypto or even when you talk about Web2 as well. I think the important thing is leaving a lot of the flexibility and decision-making in the hands of users and developers."
Developer playbook: build for agents and scale
- Design for agent payments: implement X402-compatible flows, idempotency keys, and clear retry rules. Keep gas budgeting predictable.
- Expose clean data for MCP: stable schemas, consistent event topics, and pagination for high-volume reads.
- Harden safety rails: pre-trade checks, circuit breakers, allowlists/denylists, and bounded position sizes for autonomous execution.
- Monitor and simulate: add telemetry for agent behavior, rate-limit by wallet or intent, and run chaos tests for burst traffic and partial failures.
- Plan crypto agility: abstract signing and key storage so you can rotate schemes when post-quantum options mature.
Why this matters for your roadmap
If your product depends on low latency, high throughput, or autonomous flows, you need a chain that won't stall when usage spikes. You also need protocols that agents can use natively, with clear paths to future security upgrades without breaking your UX.
That's the bet Sei is making: high-speed execution today, agent-first integrations via MCP/X402, and a realistic, staged approach to quantum-safe cryptography.
Level up your skills
If you're skilling up on agent frameworks, on-chain automation, or AI x dev workflows, browse curated tracks for developers at Complete AI Training.
Disclaimer: The information in this article is for informational purposes only and is not financial advice. We do not guarantee completeness or accuracy. All investments carry risk, and past performance is not indicative of future results. Consult a financial advisor before making investment decisions.
Your membership also unlocks:
