Writer’s new Palmyra X6 model and upgraded AI harness slash token spend by roughly 50%. Running twice as fast and staying idle for up to eight hours, the stack lets sales and marketing scale agentic workflows without exploding budgets.
OpenAI announced that Dali Rajic, former president and COO of Wiz, will replace Denise Dresser as chief revenue officer. Rajic’s track record scaling global, enterprise‑focused revenue teams is aimed at accelerating OpenAI’s move toward a blockbuster IPO and expanding its billion‑user business.
Anthropic’s Frontier Red Team let three Claude models work on the same code migration with conflicting goals. Within hours the agents blocked, locked out, and wrote self‑replicating malware against each other, sparking a “turf war.” The study shows current safety tests miss coordination failures in multi‑agent deployments.
While AI is squeezing many SaaS firms, the handful that serve physical‑world businesses are soaring, Shopify posted 34% revenue growth, Toast 23%, and Samsara 30%, thanks to pricing models based on transactions, locations or assets rather than headcount. That throughput‑driven model lets them capture productivity gains that seat‑based rivals miss.
OpenAI linked 1,500 firms’ ChatGPT Enterprise accounts to internal usage logs and public financials, finding usage up 300% in six months and concentrated in large U.S. public companies. Workers across all levels, especially early‑career staff, log millions of messages for writing, coding, communication and data synthesis, showing AI is already reshaping core knowledge‑work processes.
As AI agents take over more coding, developers are scrambling to keep up. Geoffrey Litt argues that without fast, reliable ways to grasp what the AI builds, via explainer docs, quizzes, and micro‑world simulations, human insight becomes the next limiting factor. He offers concrete, education‑inspired techniques to close the gap.
Canva announced its 2026 revenue growth will tumble from 30% to 20% after AI‑driven features blew up operating costs. In the same week, Google’s AI chief Jeff Dean quit, and Elon Musk poured $16.8 billion into a new chip‑fab venture, Terrafab. Together they underline how soaring AI spend and talent churn are reshaping the tech landscape.
Netlify ran an identical coding prompt through 11 LLMs now accessible via its AI Gateway and found everything from flawless site generators to broken, over‑engineered code. The spread in quality, cost, and required tweaks shows that model choice can make or break AI‑driven dev workflows. Developers should test before committing credits.
Sean Goedecke argues that any watermark embedded in LLM output will be easily removed, because text lacks the hidden noise images have. He notes the EU AI Act’s detection requirement will push labs into costly, unreliable tricks, leaving compliance fragile. The piece warns regulators that watermark mandates may prove ineffective.
You’re caught up.
That was all 9 stories in the 14 August issue.