- Everyday AI
- Posts
- Ep 778: Codex Goes Remote Control, claude Goes Small, notebookLM gets super Powers and 7 more AI features you Can’t skip out on
Ep 778: Codex Goes Remote Control, claude Goes Small, notebookLM gets super Powers and 7 more AI features you Can’t skip out on
Codex mobile control launches in ChatGPT, U.S. and China open high-stakes AI talks, Musk-OpenAI trial deliberations start and more
Everyday AI
May 15, 2026
👉 Subscribe Here | 🗣 Hire Us To Speak | 🤝 Partner with Us | 🤖 Grow with GenAI
In Partnership With Adobe
Adobe just introduced an entirely new way to create with Firefly AI Assistant.
Adobe just introduced an entirely new way to create — bringing the power and precision of its creative suite into one conversational experience. With Firefly AI Assistant — now live in the Adobe Firefly app, the all-in-one creative AI studio - just describe what you want to create, and the assistant orchestrates the workflows, drawing on 60+ pro-grade tools across Adobe Creative Cloud apps, including Photoshop, Illustrator, Premiere, Lightroom and more — to help bring your ideas to life.
Every step the assistant takes is visible, so you can refine, redirect, or take over at any time. You stay in control of the outcome as the creative director. Check it out today at firefly.adobe.com.
Outsmart The Future
Today in Everyday AI
8 minute read
🎙 Daily Podcast Episode: Codex remote control in ChatGPT, Claude Code gets goals, Microsoft makes big Copilot move Edge and more. Give today’s show a watch/read/listen.
🕵️♂️ Fresh Finds: Figure livestream watching humanoids work goes viral, Hermes Agent and Codex pair up, Anthropic warns of China's AI influence and more. Read on for Fresh Finds.
🗞 Byte Sized Daily AI News: Codex mobile control launches in ChatGPT, U.S. and China open high-stakes AI talks, Musk-OpenAI trial deliberations start and more. Read on for Byte Sized News.
💪 Leverage AI: It’s impossible to keep up with the new AI features that define your workflow. We do it for you, and give you the how-to guide in our newish Friday Features segment. Keep reading for that!
↩️ Don’t miss out: Miss our last newsletter? We covered: Google is teasing a new Gemini Spark agent platform, Apple is scrambling to figure out how AI agents fit into the App Store, and OpenAI wants a global AI watchdog that could even include China, and more. Check it here!
Ep 778: Codex Goes Remote Control, claude Goes Small, notebookLM gets super Powers and 7 more AI features you Can’t skip out on
The smallest AI update that'll change your work?
ChatGPT's new Codex remote control update. 🤯
But that's not the only big feature update you mighta missed this week. From Claude making a big pivot to small businesses to NotebookLM getting super powers, there's a handful and a half of AI updates this week you can't afford to not use.
In our newest 'Friday Features' segment, we break down the newest AI releases and how to actually use them.
Also on the pod today:
• Codex remote control from phone 📱
• NotebookLM powers Google automations 📓
• Goal mode hits Claude Code 🎯
Listen on our site:
Click to listen
Subscribe and listen on your favorite podcast platform
Listen on:
Here’s our favorite AI finds from across the web:
New AI Tool Spotlight – HasData is a web scraping service for AI, OpenHuman is an open source AI harness built for humans, Lokuma is an AI-native website builder
Grok Coding — Grok Build launches in early beta: a terminal-first coding agent with parallel subagents
AI Chips — The U.S. quietly cleared Alibaba, Tencent, and others to buy Nvidia’s H200 AI chips, but none have shipped yet. Find out more.
AI Robotics — Millions have tuned in to watch a livestream of Figure humanoids sort packages.
Hermes AI Agent — Hermes Agents can now run on the Codex runtime when using OpenAI models, just by flipping a switch.
AI Safety — A top AISI research leader is heading back to the Bay Area to launch a new AI alignment nonprofit.
AI Policy — In a new policy paper, Anthropic says the AI race hits a tipping point by 2028, either the US locks in its compute lead or China reaches near parity.
1. Codex Comes to ChatGPT Mobile for Live, On‑the‑Go Work Control 📲
OpenAI has pushed Codex into the ChatGPT mobile app, letting users oversee and steer active coding and research work from their phones as it runs across laptops, dev machines, or remote environments.
The move reflects how AI agents are now handling longer tasks, making quick human check‑ins more valuable for keeping work accurate and moving forward. From mobile, users can review results, approve actions, change direction, and start new threads while all files and credentials stay on the original machine.
2. U.S. and China Open AI Talks as Washington Claims the Lead 🇨🇳
U.S. Treasury Secretary Scott Bessent said Washington and Beijing have begun formal talks on artificial intelligence governance, arguing the U.S. can lead the conversation because it is ahead in AI development, according to CNBC.
The discussions come during President Trump’s high-stakes Beijing summit with Xi Jinping and focus on setting shared rules to prevent advanced AI models from falling into the hands of nonstate actors, signaling a rare point of cooperation amid deep rivalry.
3. Cerebras IPO Pops, Then Stumbles as Wall Street Weighs AI Hype 📉
Fresh off its Nasdaq debut, Cerebras Systems saw shares drop 10% Friday after a blockbuster IPO that briefly sent the stock soaring 68% and valued the AI chipmaker at about $95 billion.
The company raised $5.55 billion in the largest U.S. tech IPO since 2019, pitching its wafer-scale processors as a faster alternative to Nvidia GPUs, especially for AI inference. Some analysts are already urging caution, calling Cerebras’ technology powerful but narrow and still early in its commercial life.
4. Jury Begins Deliberations in Musk–OpenAI Trial Over For‑Profit Shift ⚖️
The first phase of Elon Musk’s lawsuit against OpenAI wrapped up this week in federal court in Oakland, putting the fate of Sam Altman and OpenAI’s for‑profit conversion in the hands of an advisory jury that starts deliberations Monday. Musk argues OpenAI broke its nonprofit promises and enriched insiders and Microsoft, while OpenAI says no such commitments existed and calls the case sour grapes after Musk launched rival startup xAI.
The jury’s verdict will guide Judge Yvonne Gonzalez Rogers, who will ultimately decide liability and, if needed, move into a remedies phase that could include leadership changes or unwinding OpenAI’s recapitalization.
5. AOC and Sanders propose national moratorium on new U.S. data centers over AI energy costs ⚡
Progressive lawmakers moved this week to pause all new AI data center construction until Congress sets nationwide safeguards for electricity prices, workers, and environmental impacts tied to AI growth.
The bill from Rep. Alexandria Ocasio-Cortez and Sen. Bernie Sanders comes as U.S. power demand hits record levels and local opposition grows over rising utility bills and water use. Though unlikely to pass, the proposal exposes a sharp divide within Washington between those urging caution on AI’s rapid expansion and leaders who see any slowdown as a competitive risk against China.
Claude built the agentic desktop. OpenAI just made it mobile.
How?
This week, Codex landed inside the ChatGPT app so your phone can now control any machine running Codex, with terminal output, diffs, and approvals streaming in real time. And it works on all plans, including free.
Meanwhile, Anthropic made a real play for small businesses, Google snuck in the most underrated automation upgrade of the week, and OpenAI's voice AI finally got GPT-5 class reasoning.
If your team is still babysitting a laptop to keep agents running, y'all are prolly already getting lapped.
On today's Everyday AI Friday Features show, we broke down all seven. Here's what you missed.
Let's get into it y’all.
Anthropic built their first SMB package as a toggle inside Claude CoWork. Flip it on and Claude connects to QuickBooks, PayPal, HubSpot, Canva, DocuSign, Google Workspace, and Microsoft 365.
It ships with 15 ready-to-run workflows and 15 skills covering payroll, invoicing, lead triage, contract review, and tax prep. Every action requires your approval before anything sends, posts, or pays.
The catch? Base plan users will burn through rate limits fast. These agentic flows aren't always token-efficient.
Try This
Toggle on Claude for Small Business in CoWork and connect QuickBooks and PayPal first.
Run the month-end close workflow before your next accountant call. If it surfaces something worth catching, stack the invoice chaser on top.
2. NotebookLM Becomes Your Automation Brain 🧠
Google dropped an Ask NotebookLM step inside Google Workspace Studio, and almost nobody covered it.
Before this, automation flows pulled from Gemini's training data or the open web. Now you can lock any flow to a specific NotebookLM notebook as the only source.
Your automations generate responses grounded only in the research and documents you control. NotebookLM's grounding is still the best in the space.
Rolling out to Workspace orgs now, with full visibility expected by end of June.
Try This
Ask your IT admin if Workspace Studio is enabled. Build a three-step flow: email trigger, Ask NotebookLM step pointed at your FAQ notebook, auto-draft reply.
If it handles repetitive questions without touching the open web, scale it across the team.
3. OpenAI Voice AI Finally Learned to Think 🎙️
GPT-Realtime-2 is OpenAI's first voice model with GPT-5 class reasoning, and it changes what voice agents can actually do.
The old approach required stitching together separate models for transcription, reasoning, and synthesis. GPT-Realtime-2 collapses that ENTIRE stack into a single audio-in, audio-out model with a 128,000 token context window, four times its predecessor.
Two companions shipped alongside it: GPT-Realtime-Translate for 70-plus input languages and 13 output languages, and GPT-Realtime-Whisper for streaming transcription. All three are live in the API now, coming to ChatGPT soon.
Try This
Test GPT-Realtime-2 in the OpenAI Realtime API playground on your gnarliest voice use case.
If it handles what your current stitched stack can't, consolidating to one model cuts overhead significantly.
4. Codex Connects Directly to the OpenAI Platform ⚡
OpenAI shipped an OpenAI Developers plugin for Codex, and it removes the most annoying part of building on the OpenAI API.
Before this, you were bouncing between Codex and platform.openai.com, wiring up API keys manually, and hunting down error messages across windows. The plugin handles all of it from inside Codex: auto-creates API keys, diagnoses errors, and works alongside the bundled OpenAI Docs skill.
Anyone on the Codex app can install it. Search OpenAI Developers in Plugins.
Try This
Install the OpenAI Developers plugin, then describe an automation you've been putting off.
Point Codex at the GPT-Realtime-2 page and say build me a prototype. Watch how far it gets before you have to touch anything.
5. Microsoft Edge Folds Copilot Mode Into Everything 🌐
Microsoft retired Copilot Mode in Edge and baked those features directly into the default browser on desktop and mobile.
Mobile gets multi-tab reasoning for the first time, plus Journeys (long-term memory) and Vision and Voice. New desktop additions include Study and Learn mode, a Writing Assistant, and a tabs-to-podcast feature.
Edge is built on Chromium, so your Chrome extensions and data sync over cleanly.
Try This
Update Edge on your phone and open five competitor pages.
Ask Copilot to compare them across all tabs at once. If it saves the tab-switching grind, it may be worth making Edge your default mobile browser.
6. Claude Code Ships a /goal Command 🎯
Instead of a prompt, you now give Claude Code a condition to meet.
After every turn, a small evaluator model (Claude Haiku by default) decides yes or no on whether the condition holds.
If yes, it stops. If no, it keeps going.
This separates the worker from the judge. Available now in the CLI for paid users on the latest version.
Desktop app still pending, and Codex had this first.
Early reception has been mixed, and a number of users called it clunky.
Try This
Open a Claude Code CLI session and try /goal with a testable condition: all unit tests pass and the linter returns zero errors.
If it completes without a single check-in, start mapping your repeating tasks to condition-based handoffs.
7. Codex Mobile Lets Your Phone Run the Show 📱
This is the BIG one.
Codex is now inside the ChatGPT mobile app on iOS and Android, and your phone can connect to and control any machine running Codex. Terminal output, diffs, and approvals all stream back in real time.
You can control multiple machines from the same account. Files and credentials stay on the host.
Claude's Dispatch connects to CoWork only and has had persistent issues. Codex is one unified interface with memory across sessions.
Available NOW on all plans, including free. Windows support coming soon.
Try This
Update both the ChatGPT mobile app and the Codex app on your Mac and pair them via the QR code.
Start a long task on your laptop, then approve it from your phone while you're out. If you've been leaving a laptop cracked open to babysit an agent, that habit ends today.
Reply