Good Morning! Here's what I have for you in today's newsletter:
OpenAI launches ChatGPT Images 2.6 with GIFs and reference images
Google adds Voice Remixing to Gemini TTS, voices tuned by prompt
xAI lets Grok Bot make phone calls and handle pushback for you
Turn ChatGPT into your photo editor with custom shortcuts
4 new AI tools worth trying today
TOGETHER WITH ATTIO
Attio is the #1 fastest-growing CRM on Brex and Ramp.
Built for the most ambitious companies in GTM, Attio is an end to end revenue platform that allows teams to run multiple plays across different revenue motions and lifecycles.
Teams at the frontier are already building on it: Granola, turbopuffer, Modal, and many many more.
AI MODELS
OpenAI released ChatGPT Images 2.6 with animated GIFs and saved reference images. It’s available on every plan in ChatGPT and Codex, across web, desktop, and mobile.
Create a GIF from a prompt or image, adjust the speed, and edit individual frames before turning it into a loop.
Save pictures of yourself, pets, or characters for future requests without uploading them again. Image markup and SVG exports add options for editing and using the results.
Two new API models offer 50% faster image generation. Flare can produce an image in as little as six seconds, while Sunburst focuses on consistency across edits.
A recurring character shouldn’t need a fresh introduction every time you make an image. Saved references cut that repetition, while GIF editing adds a way to animate results without moving to another app. Developers get two API options, though the fastest generation time won’t tell them how either model handles their particular workload.
AI TOOLS
Google released Voice Remixing for Gemini 3.8 Flash TTS in AI Studio and the Gemini API. It lets users adjust one of more than 2,000 library voices or a custom-designed voice through written prompts.
Describe what should change, from a subtle Southern US accent to the sound of a fire-breathing dragon. Instructions can alter timbre, pitch, pace, and accent.
Select “Remix this voice” in AI Studio to compare three generated takes. Choose the closest match and continue adjusting it.
Save a version for later scripts or multi-speaker scenes, keeping a recurring character’s voice available across projects.
Sometimes a voice is almost right, except it speaks too quickly or the pitch feels off. Remixing lets users adjust those qualities without starting the search over. Comparing three takes makes the differences easier to hear, and saved versions provide a consistent starting point for the next recording.
AI AGENTS
xAI rolled out phone calls in Grok Bot, allowing it to contact a business, speak with whoever answers, and bring the outcome back to your chat.
The Bot can suggest a call, such as canceling an internet contract before renewal. It presents a call card and waits for confirmation before dialing.
A live transcript shows both sides of the conversation. “Take over” lets you join, while “End call” stops it.
In xAI’s example, the Bot followed instructions to reject discounts, declined a 10% retention offer, and canceled the plan. The recording and transcript remained in the chat.
Canceling a subscription can turn into a conversation about why you should keep it. Grok Bot can take that call with your instructions, while the transcript and takeover button let you follow along. The demonstration shows it handling pushback, but xAI hasn’t specified which plans include calling, leaving availability unclear.
HOW TO AI
The photo codes you see on Reels aren’t hidden ChatGPT commands. You can define your own in a project by pairing each code with an editing instruction. You’ll need image editing available in your ChatGPT account and a few photos you took yourself.
Step 1: Create a project
Create a ChatGPT project called Photo Codes. Save your code definitions in its instructions and use chats inside that project for your edits. This gives ChatGPT the instructions behind each shortcut without making you paste them every time.
Step 2: Paste your codes
Paste the instruction block from the full guide into the project’s Instructions. It defines six codes: /studio for lighting, /cleanup for clutter, /lock for a specific change while preserving the subject, /pose for pose changes, /upscale for a sharper redraw, and /codes for the list.
Save, start a new project chat, and send /codes. Check that all six appear without an image being generated.
Step 3: Run one code
Upload a photo with a lighting problem and send /studio. If typing / opens a menu, dismiss it and check that your code remains.
Compare the result with the original. Look for improved lighting without changes to the scene or framing. If too much changed, reply: “Keep the original scene and only fix the light.”
Step 4: Stack and steer
Add details after a code, such as /pose sitting up, looking straight at the camera. Combine requests with /cleanup /studio to ask for cleanup and lighting adjustments together.
Check that both changes happened. Compare markings, textures, eyes, and text with the original. Start each fresh attempt from the original photo to reduce accumulated changes.
P.S. You can access all my AI courses, trainings, prompts library, and $500+ worth of AI tool discounts here.
Anthropic updated Claude Code to reduce usage spent on background subagent updates. Haiku 5.5 now writes them from short notes. The company says the change leaves room for about 6% more prompts within the same limits on Pro, Max, Ultra, Team, and Enterprise.
Notion launched shared memory for agents, allowing ChatGPT, Claude, Muse, and Bot to save context in one place and use the same skills. Setup begins with a Notion prompt that tells the agent how to retain information from conversations.
Cognition introduced SWE-2 Flash, a coding model built on Thinking Machines’ Inkling-Small. The company reports three times the speed at one-fifth the cost of SWE-2. Weights are planned for selected partners and researchers, rather than a general release.
📱 FlutterFlow: Launched an agent for building native mobile apps from prompts, covering interfaces, backends, authentication, and submission to the App Store and Google Play.
🎬 HyperFrames: Updated its open-source video skills with /reverse-any-video, plus music, sound effects, stock images, and TTS available free after signing in.
🖌️ Comfy Cloud: Added an official ChatGPT plugin for running cloud-hosted ComfyUI workflows from a conversation.
💬 ElevenLabs: Added mobile video captions with styles, text editing, and translation that preserves the original audio.
THAT’S IT FOR TODAY
Thanks for making it to the end! I put my heart into every email I send, I hope you are enjoying it. Let me know your thoughts so I can make the next one even better!
See you tomorrow :)
- Dr. Alvaro Cintas
✓ Full archive of premium guides with ready-to-use prompts
✓ Structured AI courses (step-by-step, start-to-finish)
✓ Every upcoming premium tutorial
✓ $500+ worth of AI tools