the brief

Agents got more portable and programmable: OpenAI’s ChatGPT app now steers Codex remotely, and Codex gained hooks for safer, automatable loops. xAI’s Grok Build debuted as an agentic CLI, while Vercel added natural‑language firewall rules and IBM shipped multilingual embeddings. Anthropic mixed policy and impact with a U.S.–China paper and a $200M initiative, as auditors flagged AI scribes’ reliability gaps.

the poursit · sip · 10 items

pulse

(06)
  • openai/blog· feedMay 14, 01:00 PM

    ChatGPT app controls Codex remotely

    Use the ChatGPT mobile app to monitor, steer, and approve Codex tasks while they run on your desktop or devbox, bringing real-time agent control across devices.

    Work with Codex from anywhere — Use Codex anywhere with the ChatGPT mobile app. Monitor, steer, and approve coding tasks in real time across devices and remote environments.

    signal 6hype 3product_updatemobile_appcoding_agentssource ↗
  • @OpenAIDevsMay 14, 09:06 PM

    Codex adds customizable automation hooks

    New hooks let you run scripts before and after work, scan prompts for secrets, log conversations, and build memories—making the Codex loop safer, auditable, and extensible.

    Codex is getting easier to automate and customize around your code. 🪝 Hooks customize the Codex loop with scripts that run at key points in a task: • Run validators before or after work • Scan prompts for secrets • Log conversations to internal systems • Create memories or pic.x.com/mb0LhxwHR6

  • @vercel_devMay 14, 10:00 PM

    Vercel Firewall gets natural-language rules

    You can now define firewall rules like path blocks, rate limits, and challenges in plain English, speeding up protection without detouring into regexes or config syntax.

    Vercel Firewall now supports creating custom rules with natural language: ▪︎ Block all /wp-admin requests ▪︎ Rate limit /login to 5 attempts per IP ▪︎ Challenge /checkout requests from node vercel.com/changelog/crea…

    signal 6hype 2product_updatevercelfirewallsource ↗
  • @unknownMay 14, 06:32 PM

    Grok Build agentic CLI beta

    Early beta of Grok Build offers an agentic CLI for coding, app building, and workflow automation to SuperGrok Heavy subscribers, with feedback driving rapid iteration.

    An early beta of Grok Build, an agentic CLI for coding, building apps, and automating workflows is now available for SuperGrok Heavy subscribers. Through this early beta, we will improve the model and product based on your feedback. Try it at x.ai/cli pic.x.com/Rlg4qMLkrv

    signal 7hype 2agent_cliproduct_launchagent_frameworksource ↗
  • @AnthropicAIMay 14, 03:08 PM

    Anthropic, Gates launch $200M initiative

    Anthropic and the Gates Foundation pledged $200M in grants, Claude credits, and technical support to accelerate AI applications in health, education, agriculture, and economic mobility.

    We’re partnering with the Gates Foundation, committing $200 million in grants, Claude credits, and technical support to programs in global health, life sciences, education, agriculture, and economic mobility. Read more: anthropic.com/news/gates-fou…

  • huggingface/blog· feedMay 14, 06:55 PM

    Open multilingual embeddings with 32K context

    IBM’s Granite Embedding Multilingual R2 lands under Apache 2.0 with a 32K context window and state‑of‑the‑art sub‑100M retrieval performance for multilingual RAG.

    Granite Embedding Multilingual R2: Open Apache 2.0 Multilingual Embeddings with 32K Context — Best Sub-100M Retrieval Quality

    signal 8hype 2model_releaseembeddingsmultilingualsource ↗

findings

(02)

voices

(02)
  • @ClaudeDevsMay 14, 11:35 PM

    Pre‑warm Claude’s prompt cache

    Send your system prompt first to seed the cache without generating output, cutting time‑to‑first‑token for long prompts in production API workloads.

    Useful tip to cut time-to-first-token on longer prompts in the API: pre-warm the prompt cache. Send your system prompt before the user prompt. Claude writes it to the cache, but skips generating any output. When the real user request lands, it'll hit a warm cache. pic.x.com/6BdEzbamr2

    signal 9hype 1prompt_cachelatency_optimizationclaude_apisource ↗
  • @gdbMay 14, 09:15 PM

    Mobile Codex unlocks ubiquitous agents

    Greg Brockman frames using Codex from the ChatGPT app as a major step toward agents that work continuously and follow you seamlessly across devices.

    You can now use Codex, wherever you have it running, from the ChatGPT app. Huge step forward for universal usage of agents. x.com/openai/status/…

    signal 7hype 3agent_platformchatgptcodexsource ↗