Sponsored by
👀 AI Signal Overview
Inside this issue | What you’ll learn |
|---|---|
🛠️ Hack of the Week | One global instruction that makes AI sound 80% less like AI |
⚡ Three signals worth knowing | Meta’s local AI agent, Anthropic’s reported $9.1B infrastructure deal, and unlimited ChatGPT Free text chats. |
📬 AI hiding inside your inbox | Summarize Apple Mail threads, track Yahoo orders, use Copilot with Outlook, and ask Gmail questions in plain English. |
💸 Before you renew another subscription | Four free Google Labs tools covering social content, interface design, automations, and visual concepting. |
🛠️ HACK OF THE WEEK
AI Models are increasingly talking to us using words and jargon we don't understand.
One verified fix for this is to set the following global instruction...
"Use ASD-STE100 Simplified Technical English for all responses"
You can use it in ChatGPT, Claude, Gemini, Grok, Perplexity, and Cursor. It is a writing instruction, so it is not tied to one model.
Use it for emails, instructions, SOPs, reports, documentation, and explanations. Avoid it when you want personality, storytelling, humor, or creative writing—it can make those feel flat.
The best voice models now listen, adapt, and resolve too.
Most CX platforms don't own the voice. They orchestrate a workflow, then call a third party for speech and transcription. Every hop adds latency, and latency is what turns a frustrated customer into a churned one.
ElevenAgents is the opposite. Built on the voice models the market already builds on, it runs voice, transcription, chat, and reasoning in one vertically integrated pipeline. Responses come back in under 400 milliseconds and sound human, not synthetic. When a caller gets frustrated, the agent detects it and shifts tone in real time: calm, reassuring, patient.
You keep full control. Plug in any LLM, connect tools, webhooks, and MCP servers, and ground every answer in your knowledge base. Launch in minutes, A/B test with Experiments, enforce Guardrails, and version every change.
More resolved conversations, less infrastructure stitching. Pricing is transparent and flat at $0.08 per minute.
⚡ Three signals worth knowing
Meta released Muse Glimmer, a 30-billion-parameter open-weight model built to run agentic tasks locally on a Mac or PC with a single graphics card. It is Meta’s answer to increasingly capable Chinese open models from DeepSeek, Alibaba, and Moonshot.
Why it matters: Your files can stay on your device, and developers can customize the model without paying a closed AI provider for every request. Local AI is becoming a real product category, not just a hobbyist experiment.
Riot Platforms signed a 20-year lease for 191 megawatts of AI data-center capacity at its Rockdale, Texas campus. Riot did not name the customer, but Bloomberg identified it as Anthropic.
The base contract runs through June 2048 and could reach $16.1 billion if both five-year extensions are used.
Why it matters: The AI race is no longer measured only by model benchmarks. Labs are locking up electricity and data-center space decades in advance—and Anthropic is preparing to operate at a much larger scale.
OpenAI is making GPT‑5.6 Luna the default model for Free and Go users. It is also rolling out unlimited text chats and a new Think button for harder questions.
“Unlimited” applies to text chats and remains subject to abuse guardrails. Separate limits still apply to images, file uploads, and other tools.
Why it matters: The basic chatbot is becoming nearly free and unlimited. The paid advantage is shifting toward stronger reasoning, higher tool limits, and AI that can complete work—not merely answer questions.
Google spent 10 months shipping the individual jobs your subscriptions do, as separate free products.
They live at 5 different URLs under Google Labs. That's the whole reason the total is easy to miss. Each one looks small on its own. Added up, they cover most of what you pay for.
So the question worth 10 minutes: what's still on your card that shouldn't be
The 5 tools, and what each one covers.
1. Pomelli. Give it your website URL. It reads your site and builds a profile Google calls Business DNA: your colors, your fonts, your tone of voice. Then it writes social posts and generates product photos from that profile. Free, beta, any Google account.
Covers: the reason you have Canva Pro.
2. Stitch. Describe a screen in plain English. It generates the interface and the code, then exports to Figma or as HTML and CSS. No paid plans exist yet.
Covers: the Figma seat you open twice a month.
3. Opal. Describe an automation in one sentence. It builds a visual workflow you can edit, and outputs land in Docs, Sheets, Slides and Drive. Free public beta, 160+ countries.
Covers: the Zapier or Make plan you set up once and never grew.
4. Mixboard. An AI concepting board. Drop in text and images, generate more with Nano Banana, then keep editing in plain language until it looks right. Free beta.
Covers: the 40 minutes you spend hunting reference images before you make anything.
⚠️ Read this before you cancel anything.
All 4 are Google Labs experiments. Labs pricing is a snapshot, not a promise.
So here's the rule I'd use. Move the work you can afford to lose. Social graphics, mockups, moodboards, throwaway automations: move those today. The client deliverable due Thursday, the file your team edits, the automation that breaks something if it stops quietly: leave those alone. A free tool that gets quota-cut mid-project isn't free. It moved the cost onto your Thursday.
Until next Tuesday,
Dan Rice from AI Signal