VentureBeat

Autonomous agents make decisions. Most security frameworks weren't built for that. Amy Chang, Head of AI Threat Intelligence and Security Research at Cisco, has been mapping the agentic threat surface that most enterprise teams haven't looked at yet, prompt injection, non-human identity management, runtime policy gaps, and governance in systems that don't wait for human approval. She's on the AI security panel at VB Transform 2026. If you're architecting agentic systems, this is the conversation to have before something goes wrong. July 14–15 | Hotel Nia, Menlo Park 🔗 https://bit.ly/49oTA6r

Valid certificates, stolen accounts: how attackers broke npm's last trust signal https://bit.ly/4v3OvcG

AI writes the code. Engineers inherit the consequences. Jared Go,Distinguished Engineer at Symbotic, author of AI Coding: Beyond the Vibe is joining VB Transform 2026 to talk about what actually comes after the prototype: the architectural decisions, the failure modes, and what it takes to ship AI-generated code you'd actually trust in production. July 14–15 | Hotel Nia, Menlo Park Meet him there: https://bit.ly/497VGZt

What does it mean to present at the VB Transform Innovation Showcase? You get on stage in front of hundreds of enterprise AI decision-makers. You get direct feedback from a curated panel of thought leaders. And you get exclusive editorial coverage from VentureBeat — in front of millions of monthly readers. Applications are open through June 1. https://lnkd.in/gsFC2Fpa

VentureBeat's Q1 data shows enterprise AI just hit a wall: cost-per-inference anxiety jumped from 34% to 41% in one quarter while GPU panic dropped. The scramble is over. The audit has begun. But most enterprises aren't ready for it — 72% admit they lack sufficient control over their AI infrastructure. LinkedIn CTO Erran Berger built the answer from the ground up: bare metal, custom GPU kernels, model pruning, embedding compression. The full stack engineered for ROI — not just throughput. What does it actually take to get there? https://lnkd.in/gusYSDne

D&B's database of 642 million businesses was built for humans, not AI agents. So they rebuilt it. https://bit.ly/4fBzSJd

Alibaba's proprietary Qwen3.7-Max can run for 35 hours autonomously and supports external harnesses like Anthropic's Claude Code https://bit.ly/42LpyH9

What does it actually mean to have organizational discipline around AI ROI? Not just dashboards — the mindset that every feature needs an ROI analysis before it ramps to 100% production and the infrastructure to make that analysis possible. LinkedIn CTO Erran Berger built that culture. Here's what it looks like in practice. https://lnkd.in/gusYSDne

A 0.12% parameter add-on gives AI agents the working memory RAG can't https://bit.ly/4fBLjR3

Last night, we brought together AI experts in NYC for a candid conversation on the production reality check: Reliability - How to Engineer AI Agents That Actually Ship. The room was energized. Questions flew. And one thing became clear: it's not the models holding teams back—it's orchestration and workforce reliability. VentureBeat CEO Matt Marshall shared early findings from our ongoing research into what AI leaders actually struggle with when determining if an agent is production-ready. (Hint: the answer wasn't the model.) Key insights from our panelists: Andrew Reiskind, Chief Data Officer at Mastercard, discussed engineering trust levels—building guardrails for various agents and setting thresholds to determine when they're ready to go live. Sean Finnerty, VP of Digital Platforms & Services at Merck, reminded us there are no shortcuts: orchestrating drug discovery agents or navigating global pharma marketing regulations means wrestling with an enormous variable set. The tech alone won't solve it. Preeti Somal, SVP of Engineering at Temporal Technologies, explained how resilient orchestration manages long-running workflows, handles API failures, and keeps conversation states intact—so your processes don't break when the inevitable errors hit. One theme echoed throughout: the "DIY tax" teams are paying to plumb together their agentic workforce is real—and expensive. We host these intimate, expert-led sessions throughout the year in NYC and San Francisco. Space is limited. Want in? Join our newsletter to stay in the loop on upcoming events in your area.