- AI Toast
- Posts
- Gemini AI is now inside your TV
In partnership with
Hey there, and welcome to AI Toast!
Today’s menu:
Gemini comes to Google TV
OpenAI x Nvidia
Notion’s new AI Agents will do your job for you
Superhero Tools (Coding & Vibe Coding)
How to build your first AI Agent
Quick AI Toast: Facebook is fighting to get you a love life
Total read time: About 7 minutes, perfect for a quick cappuccino.
Gemini comes to Google TV
Image Source: Google
Google is bringing its Gemini AI to Google TV, aiming to transform how you interact with your television. This advanced language model makes TV smarter, more intuitive, and highly personalized.
Gemini goes beyond voice commands: have free-flowing chats about what to watch, brainstorm family trips, or get homework help.
Just say “Hey Google” or tap your remote to start a conversation with your TV.
Find tailored content by describing your mood or companions (“What’s good for movie night if I like drama but my wife likes comedy?”).
Get quick recaps by asking about your favorite show’s latest season, or let Gemini guess that “new hospital drama” everyone's buzzing about.
Ask follow-up questions, check reviews, and dig deeper — no need to switch devices.
Learning gets interactive: Gemini answers school project questions, shows YouTube tutorials, guides your kitchen experiments, or teaches new skills right on the TV.
Launching on TCL QM9K now, with more TVs and streamers in the coming months, 300 million active Google TV/Android TV OS devices worldwide.
Gemini on TV signals an era where our biggest screens become truly smart companions. Expect your living room experience to grow more natural and helpful.
Personal Take: This feels like the start of AI becoming the heart of home entertainment. Conversation-powered discovery, learning, and leisure, all from your sofa, guided by true artificial intelligence.
Smart dictation that understands you
Typeless turns your raw, unfiltered voice into beautifully polished writing - in real time.
It works like magic, feels like cheating, and allows your thoughts to flow more freely than ever before.
With Typeless, you become more creative. More inspired. And more in-tune with your own ideas.
Your voice is your strength. Typeless turns it into a superpower.
Image Source: OpenAI Blog
The future of artificial intelligence just got a massive boost. OpenAI and NVIDIA’s multi-billion dollar partnership is laying the foundation for a new era of superintelligent compute, promising breakthrough AI capabilities at global scale.
OpenAI and NVIDIA join forces to deploy at least 10 gigawatts of NVIDIA AI datacenters, representing millions of GPUs for next-gen AI models.
NVIDIA pledges up to $100 billion investment into OpenAI, releasing funding as each gigawatt of hardware goes live.
First gigawatt launches in the second half of 2026 on NVIDIA’s Vera Rubin platform.
The collaboration co-optimizes both OpenAI’s software roadmap and NVIDIA’s hardware innovations.
This partnership builds on years of collaboration and complements joint work with Microsoft, Oracle, SoftBank, and Stargate.
OpenAI’s reach: over 700 million weekly users across businesses, enterprises, and developers worldwide.
The goal: to build superintelligence and make AI’s benefits accessible to all.
As the details finalize, one thing is clear, this is a generational leap for AI infrastructure, with ripple effects across every industry.
Personal Take: We’re witnessing the birth of an AI-powered economy. The scale and ambition here signal superintelligence isn’t a sci-fi dream, it’s a fast-approaching reality we’ll all need to adapt to.
Create How-to Videos in Seconds with AI
Stop wasting time on repetitive explanations. Guidde’s AI creates stunning video guides in seconds—11x faster.
Turn boring docs into visual masterpieces
Save hours with AI-powered automation
Share or embed your guide anywhere
How it works: Click capture on the browser extension, and Guidde auto-generates step-by-step video guides with visuals, voiceover, and a call to action.
Image Source: Notion
Notion 3.0 introduces advanced AI Agents that automate knowledge work for teams and individuals. The Agents can perform human-like tasks within Notion, including building databases and executing workflows.
Notion Agents can handle multi-step tasks, creating pages, databases, and reports, and updating them with new data.
Agents can compile customer feedback from various sources, synthesize findings, and create structured databases.
Users can trigger agents manually or schedule them to run autonomously, streamlining tasks and freeing up time for more strategic work.
Users can customize their agent's instructions and memory to fit their work style.
Notion plans to introduce custom agents that can run on autopilot, allowing users to automate specific workflows.
Agents can connect with various tools, including Slack to access relevant information.
Personal Take: While Notion's AI Agents can significantly reduce busywork, they're designed to augment human capabilities, not replace them. By automating routine tasks, users can focus on higher-level work and strategic decision-making.
Superhero Tools (Coding & Vibe-Coding)
Rocket.new: Instantly build web and mobile apps from your ideas, using AI-powered no-code tools.
Claude Code: Quickly map, explain, and analyze entire codebases using agentic AI search.
Cursor: Boost your team’s productivity and catch code bugs faster with this advanced AI coding assistant.
CodeRabbit: Get automated, AI-driven code reviews to speed up development and improve code quality.
GitHub Copilot: Write code faster as AI suggests code, functions, and entire solutions inside your favorite editor.
Replit: Instantly start coding, collaborating, and deploying software projects—no setup required.
Quick AI News Bites
Perplexity’s Comet AI browser is now available in India for Windows and Mac users. The browser helps with research and work, offering smart recommendations based on what you read.
Grok 4 Fast by xAI is a new AI model focused on cost-efficient intelligence. It delivers high-quality reasoning with faster and cheaper token usage compared to rivals, and is available for free for a limited time.
Facebook launches an AI-powered dating assistant to challenge Tinder and Hinge. This new tool uses artificial intelligence to help users match and chat more easily for dating.
Rocket.new, a Surat-based startup, raises $15M to transform app-building through “vibe coding”. The platform lets users create production-ready apps from natural language prompts and is gaining traction globally.
Build your first AI Agent
Image Source: McKinsey
Based on proven workflows and practical advice for beginners:
Choose a single, specific task you want your agent to do.
Examples:
Book a doctor’s appointment on a website
Monitor job boards and send matching jobs
Summarize unread emails
The smaller and clearer the problem, the easier it is to start and debug.
Don't waste time training your own model at the start. Use existing, reliable models: GPT, Claude, Gemini, or open-source options like LLaMA, Mistral.
Make sure the model can handle simple reasoning and structured responses.
Your agent needs to interact with the outside world.
Common actions:
Web scraping/browsing (Playwright, Puppeteer, API access)
Email and calendar APIs (Gmail API, Google Calendar)
File operations (read/write, parse PDFs, etc.)
Take user input (the goal/task). Pass it to your model with clear instructions (prompt) and let the model decide the next step
If needed, call the required tool/API. Give the result back to model for the next step.
Repeat until the task is finished or output is sent to user
Start with short-term memory (recent steps/messages only). If your agent needs to “remember” across runs, use a simple JSON file or database.
Don’t add complicated memory (vector databases, fancy retrieval) until needed.
Start with a command-line interface (CLI), just run it from your terminal.
Later, wrap it in a simple interface:
Web dashboard (Flask, FastAPI, Next.js)
Slack, Discord bot
Clickable desktop script
Run real tasks, see where it breaks, and fix bugs. Repeat this cycle until it works reliably.
Don’t keep adding features/tools. Focus on building ONE well-functioning agent for a single real-world task.
Pro Tip: Once you build one end-to-end agent, building the next one will be 10 times easier because you’ll understand the workflow!
Quick Checklist:
Pick one tiny problem.
Use an existing model.
Connect to one outside tool/API.
Build the workflow: User → Model → Tool → Model → Result.
Start in CLI, then wrap if needed.
Test and fix in cycles.
Focus on making ONE agent work well.
OpenAI also published a guide on building AI Agents. Read here
Boost revenue and gain new customers by partnering with us
Reach over 35K AI enthusiasts with your product.
Join our newsletter to connect with tech professionals, investors, engineers, managers, and business owners worldwide. DM now!
Got feedback, a story worth toasting, or a wild tech question? I’d love to hear from you; just reply, or find me on X.
Cheers,
— Poonam Soni