Vibe Coding Is Dead. Long Live Agentic Engineering.

Vibe Coding Is Dead. Long Live Agentic Engineering.

The developer discourse has quietly shifted from "vibe coding" — prompting an AI to generate code with minimal oversight — to something more structural: agentic engineering. Simon Willison and others have been mapping this convergence since May, and it's still the dominant framing in July [4]. The distinction isn't semantic. Vibe coding treats AI as an autocomplete; agentic engineering treats it as a workforce that needs contracts.

Practically, that means typed schemas (Pydantic being the recurring example), validation layers, and explicit handoff protocols between agents — because unstructured prompting doesn't scale past a demo [5]. The developers pushing this hardest aren't anti-AI; they're the ones who got burned maintaining vibe-coded systems six months after launch and learned the expensive way that "it works" and "it's maintainable" are different claims.

This is close to the exact thesis Up North AI has been building around: code is free, judgment isn't. The judgment now lives in how you orchestrate, validate, and constrain agents — not in whether you can get one to spit out a function. Voices on LinkedIn and X are increasingly blunt about this: vibe coding was a phase, not a paradigm [6].

A10 Networks Buys TrojAI — Security Catches Up to Agents

A10 Networks announced on June 15 that it's acquired TrojAI, a firm specializing in securing and governing AI applications and agentic workflows [7][8]. The deal folds AI-specific threat testing and governance into A10's existing firewall and security stack, targeting sovereign AI deployments across on-prem, cloud, and hybrid environments [9].

This is a quiet but telling acquisition. As agentic systems move from prototypes to production — handling real transactions, real customer data, real infrastructure access — the attack surface stops being theoretical. A10 isn't buying a research lab; it's buying the ability to say "we secure agents" to enterprise buyers who are asking that question for the first time and don't like the current answers.

Expect more of these bolt-on acquisitions in H2 2026. Every serious enterprise infrastructure vendor now needs an "AI security" story, and buying it is faster than building it.

Trump Pushes Clarity Act, Ties Crypto Rules to the AI Race With China

President Trump renewed pressure on the Senate to pass the Clarity Act, framing clear crypto/digital asset regulation as inseparable from U.S. competitiveness in AI against China [10]. The bill would split oversight of digital assets between financial regulators, and Trump's argument is explicitly geopolitical: regulatory delay pushes capital and talent toward competitors [11].

Democrats have pushed back hard, with Warren, Blumenthal, Durbin, and Peters criticizing Trump's own crypto-related income amid the push [11], and crypto communities on Reddit are split on whether the China framing is substantive or "copium" to force a vote [12].

The AI angle here is mostly rhetorical leverage — but it's a preview of how AI competitiveness will get invoked in adjacent policy fights all year. Expect "AI race with China" to become the go-to justification for regulatory asks that have nothing directly to do with model development.