Good Morning! Here's what I have for you in today's newsletter:
Google launches Gemini 3.8 Flash TTS, custom voices in 100+ languages
Anthropic consolidates Claude Marketplace into 2,000 connectors
Cursor launches Rollouts, a bot that watches every deploy for regressions
Get 100% out of Opus 5.5 with fixes straight from Anthropic's playbook
4 new AI tools worth trying today
AI MODELS
Google released Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS, two text-to-speech models available in the Gemini API, Google AI Studio, Gemini Enterprise, Gemini Notebook, and Google Vids.
Flash TTS builds a voice from a text description of role, accent, and character in over 100 languages and saves it for reuse, alongside a library of 2,000+ ready voices.
Voice replication recreates a speaker from a 30-second sample, gated by a separate verbal consent recording that must match the reference speaker before the voice is created.
Delivery can be directed line by line with cues like <laughs> and |mhm|, and Flash-Lite handles high-volume dubbing and voice agents at lower cost.
The clone-from-30-seconds feature is the one to test, and the consent gate means you can't quietly clone a colleague. For creators, a saved custom voice that holds up across hours of audiobook or podcast narration is the practical win. No pricing is published yet, so check AI Studio before budgeting on it.
AI TOOLS
Anthropic consolidated Claude Marketplace into one catalog of more than 2,000 connectors, plugins, agents, products, and service partners, covering discovery, purchase, and deployment help for Claude customers.
Connectors and plugins from Atlassian, Google, Microsoft, Notion, Salesforce, and Slack are built on MCP and Agent Skills, and any builder can submit one for listing.
Enterprise customers can put part of their committed Anthropic spend toward Claude-powered products from Cursor, CrowdStrike, Harvey, Legora, Lovable, and Snowflake, with Anthropic handling the invoice.
A service partner tier lists Accenture, Boston Consulting Group, and Deloitte from the Claude Partner Network for companies that want help planning and rolling out Claude across teams.
The connector directory is useful to anyone on a paid plan, and worth a browse if you haven't added Slack or Notion yet. The committed-spend part is for companies with an Anthropic contract, and Anthropic hasn't said how big a portion they can redirect. For a solo user, the directory is the whole story.
AI AGENTS
Cursor launched Rollouts, a bot for Teams and Enterprise plans that attaches a monitor to every pull request and watches the change as it deploys, alongside an upgraded Security Reviewer.
When a PR opens, Rollouts reads the diff and posts an editable monitoring plan as a comment, listing risks, intended effect, signals to check, and instrumentation gaps.
On deploy, it runs the plan against logs, metrics, and traces from Datadog, Grafana, or Honeycomb, compares them to a pre-deploy baseline, and reports verified healthy, regression detected, or inconclusive per environment.
On a regression it can ping the author, pause a progressive rollout, or open a revert PR, and it catches problems limited to one endpoint or region before a broad alert fires.
Coding agents can push more changes than a team can keep an eye on once those changes are live. Rollouts does that watching for you and flags a broken deploy before users hit it. It only works on Teams and Enterprise plans, and you have to connect your deploy system and monitoring tools first.
HOW TO AI
Opus 5.5 works longer on its own, reports plainly, and thinks before every reply with no prompting needed.
Step 1: Name the finish line
Give the whole task in one message, with a clear finish line.
Migrate the payment endpoints from the old client to the new one. Done means: every endpoint uses the new client, the old client is deleted, and the test suite passes. Stop and ask me only if a test fails for a reason you can't explain.
Step 2: Delete "think carefully"
It already thinks before every reply, the line just slows things down.
Answer directly.Step 3: Stop-and-go rules
Name the stops you want directly, or it may pause to report instead of continuing.
When a step doesn't need my input, keep going. Put status notes in the same message as your next action.
Stop and ask only when you can't continue without me, or before anything destructive: deleting data, force-pushing, or changing anything outside this repository.Step 4: Name styles to avoid
With no direction, it falls back on default styles, name the specific patterns instead.
Build a personal website with placeholder content. Don't use a cream or off-white background, italic accent words in headings, numbered "01 / 02 / 03" section labels, monospace labels, or pill-shaped buttons.Step 5: When a message gets flagged
A flagged message switches to an older model automatically.
Claude apps: choose Opus 5.5 in the model picker to go back.
Claude Code: run /model to switch back.
P.S. You can access all the AI courses and trainings (including the full version of this one), prompts and workflows if you upgrade.
Anthropic moved Claude Code cloud sessions out of research preview, so Claude keeps working after the laptop closes, and is giving existing subscribers a one-time credit to try them, $100 on Pro and $250 on Max.
Google added new creative controls to Google Vids powered by Gemini's Omni 1.1 multimodal model, and opened the AI video features to anyone with a Google account at no cost on desktop at vids.new.
ComfyUI launched Comfy Router, one API for image, video, 3D, and audio models across Comfy, Runware, Wavespeed, Fal, and Higgsfield, with explicit provider routing, no silent fallback, async batch queuing, and billing through Comfy credits with no router fee.
⚡ Recraft: added V4.1 Flash, which produces an image in 1.3 seconds from the model picker, with a Refine button to clean up details afterward.
👀 Hermes Desktop: streams any bot's screen or session live, so you can watch it drive a browser or terminal, take over to type credentials, and hand back.
🎨 LM Studio: Bionic added an interactive Excalidraw canvas that both you and the agent can edit, for mockups, system designs, and process maps it can then implement.
🎵 invideo: added Agentic Sound Design, so editing agents find or generate music and sound effects, place them on the timeline, and mix and master the sound bed.
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