Hello!

Vercel just made a big move by acquiring Better Auth, aiming to give AI agents their own identity system.

Meanwhile, Apple's tweak to Safari has opened the gates for AI agents to control the browser, making waves in the AI world. And over at Coinbase, they've managed to halve their AI expenses while running 1,200 agents—talk about efficiency!

At Pagepro, we're all about sanding down complexities with Next.js & Sanity migrations for enterprises.

Grab your coffee, settle in, and enjoy Frictionless.

No more storing tokens or secrets manually. With Vercel Connect, managing credentials for Chat SDK bots becomes a breeze.

The new @vercel/connect/chat subpath offers adapter helpers for platforms like Slack, GitHub, and Linear. This means you can smoothly integrate your existing tools without the headache of managing tokens.

Each API request now benefits from fresh, short-lived tokens that Vercel automatically rotates. This eliminates the need for cumbersome signing secrets, simplifying the security aspect of bot management. To get started, delve into the documentation or kick off a project with the Slackbot guide. Ready to simplify your bot operations?

Streamline your multi-service applications on Vercel with Service Bindings. Vercel's latest feature allows seamless and secure calls between services within the same deployment.

Once a service binding is declared, Vercel handles URL rewrites, routing, and TLS authentication behind the scenes. No need for complex public route configurations or certificate headaches. Key insight: Service isolation is maintained, yet services communicate effortlessly. Curious about implementation? Check out Vercel's documentation for a deeper dive.

AI might boost code production, but if your team’s drowning in unreviewed pull requests, it's all for nothing. Startups in hypergrowth often forget that bottlenecks aren’t about writing code — they're about getting it approved.

Faster coding doesn't equal faster shipping. Focus on building efficient review processes and empowering your team to keep things flowing. Get your team aligned on priorities and eliminate the review queue nightmare. Is your code review process up to the challenge?

Coinbase just pulled a fast one on AI costs, cutting their spend in half without breaking a sweat on token usage caps. How? Smart routing and budget-friendly defaults. Brian Armstrong's approach lets engineers choose cheaper open-weight models for most tasks, tapping high-end options only when necessary.

That's a 3-6x cost reduction per token.

Numbers don’t lie. Coinbase's cache hit rate skyrocketed from 5% thanks to task-specific routing and a strategic LLM gateway. These strategies make teams leaner and meaner, architecting AI spend without sacrificing performance.

Ready to rethink how you route AI workloads? It’s time to crush that overhead.

AI might write code, but can it replace engineers? Hrishi Dixit from Autonomy Finance dives deep into this question. In regulated industries, handing over to AI isn't just risky — it's a potential disaster.

First drafts from AI models aren't shippable, and mix-ups in client data could sink a company fast.

The engineers who'll thrive are those who can preempt AI's pitfalls. It's not about out-coding AI, but out-thinking its limitations. How are you preparing your team to navigate the AI landscape?

Codex and Claude Code are the heavyweights in AI coding assistants. These aren't your grandfather's autocompletes; they can write, debug, and even refactor. Codex operates like a development partner, managing tasks across environments seamlessly. Claude Code, meanwhile, might align better with certain team workflows through different philosophies in code reasoning and integration.

Codex shines when you need an AI that can juggle complex tasks autonomously. Workflow fit is king here. Do either of these AI tools fit your team's unique style?

AI agents need their own passports. Vercel's acquisition of Better Auth aims to solve a problem plaguing the AI ecosystem: identity confusion. When an AI acts, it's not just a shadow of its human deployer anymore.

With Better Auth's framework, AI agents get distinct identities with their own permissions, making operations more secure and streamlined. No more blanket permissions or system-wide shutdown headaches. Instead, Vercel's infrastructure promises better control and scalability.

Knowing who's who in AI isn't just neat—it's necessary. How will you leverage this new frontier in AI identity?

In the AI landscape, standards aren't just helpful—they're critical. The Linux Foundation steps up with the Agent Name Service (ANS), aiming to bring order to the wild west of AI agents.

They're also backing the DNS-AID proposal to tackle identity ambiguity. ANS means each AI agent can have a unique, standardized identity across platforms. This isn't another buzzword. It's about making AI agents more reliable and interoperable.

Today, there are 1,000+ AI agents, each with different naming conventions. Imagine them speaking the same language. That's what ANS promises CTOs: clarity and collaboration. Interoperability is the linchpin for scalable AI-driven systems. Ready to see your AI agents play well with others?

Apple's latest move: turning Safari into a playground for AI agents.

With the new MCP server in Safari Technology Preview 247, AI agents gain direct access to browser capabilities. Screenshot this, inspect that — all from your terminal. It’s like giving your AI a set of keys to Safari's kingdom. But here’s the kicker: all this happens without touching your personal data. Why does this matter?

Does this transform how you interact with AI on your platform, or does it just raise new questions?

Neuralink has done the brain equivalent of threading a needle, but without making a hole. They've managed to bypass cutting through the dura mater, a tough layer protecting our brains, by threading electrodes directly through it. It’s like deleting a surgery step altogether.

Why care? First, it's a big deal for patient safety. Less cutting means less risk. Second, it speeds up recovery time. Patients spend less time on the mend, more time back at life. Fewer steps, less risk, faster recovery.

Dare we ask — what does this mean for the future of surgical innovation?

AI hardware has been quietly brewing a revolution. June marked a turning point when custom chips hit the ground running, thanks to heavyweights like OpenAI and Amazon. It's not just lab talk anymore — these are tangible products changing the game. Forget about settling for ordinary processing power. With these new chips, AI efficiency and capabilities are skyrocketing.

Etched's custom chips promise to redefine what's possible in AI. The industry is buzzing, and for good reason. How will this shape the next wave of AI innovation? Only one way to find out. Dive into the details and see where this journey takes us.

Do you have any comments about this newsletter issue or questions you want to ask? Drop me a message or book a meeting.