Perplexity Computer, Nano Banana 2, Samsung AI Phone | Weekly Digest
+ AI Agents started layoffs & memes of the week
Hey! Welcome to the latest Creators’ AI Edition.
Google just dropped an image generator that went viral once — and now it’s back, faster and smarter. Perplexity went from “answer engine” to “digital worker” in one launch. Samsung basically declared the chatbot era over at Unpacked. Another seven days in AI, another week where everyone shipped something wild. Let’s break it all down.
Featured Materials 🎟️
News of the week 🌍
Useful tools ⚒️
Weekly Guides 📕
AI Meme of the Week 🤡
AI Tweet of the Week 🐦
(Bonus) Materials 🎁
Keep your mailbox updated with practical knowledge & key news from the AI industry!
Featured Materials 🎟️
Nano Banana 2: Google’s Viral Image Generator Is Back — Faster, Smarter, Free 🍌
Remember when Nano Banana went viral last August and millions of people were generating images in Gemini? Google just leveled up. On February 26, they launched Nano Banana 2 (officially Gemini 3.1 Flash Image) — and this time it brings Pro-quality features to everyone at Flash speed.
What’s new?
Pro features, now free: Subject consistency across up to 5 characters and 14 objects in a single workflow. Previously locked behind premium tiers.
Sub-second 4K synthesis: Resolutions from 512px to full 4K, in any aspect ratio. Vertical Stories, widescreen presentations, high-res backdrops — all production-ready.
Precision text rendering: Generate legible text inside images for marketing mockups, greeting cards, and infographics. It even translates and localizes text within images.
Real-world knowledge grounding: Pulls from Gemini’s knowledge base and real-time web search for more accurate renderings of real subjects.
Rolling out everywhere: Default in the Gemini app (Fast, Thinking, Pro modes), Google Search, AI Mode, Google Lens, Flow (video editing), AI Studio, Gemini API, Vertex AI, and Gemini CLI.
Nano Banana Pro sticks around for “high-fidelity tasks requiring maximum factual accuracy.” Google AI Pro and Ultra subscribers retain access via the three-dot menu.
See It in Action: Wow Examples 🤯
Google built two demo apps that show what Nano Banana 2 can actually do for creators:
🐾 Pet Passport: Upload one photo of your pet. Nano Banana 2 drops them into famous landmarks around the world — Eiffel Tower, Great Wall, Colosseum — while maintaining their exact appearance. Perfect for pet influencers and viral pet content.
🌍 Global Ad Localizer: Feed it an English ad. Nano Banana 2 translates the text AND adapts the visuals for different markets. Japanese version? Hindi version? Arabic? It handles the copy AND the cultural context in one shot. This alone is worth it for any creator or marketer targeting multiple regions.
Meanwhile, the viral prompts community has already gone wild. Some highlights from the first 24 hours on X:
Generating fake but photorealistic YouTube thumbnails from text-only prompts
Creating 6-panel comic stories with 3 consistent characters across every frame
Building entire product mockup photoshoots — no camera, no studio
Rendering accurate data visualizations and infographics from descriptions
Try it yourself: Open Gemini → type any image prompt. Nano Banana 2 is now the default. Or use AI Studio for developer access.
Source: Google Blog | Developer Blog | TechCrunch | BGR Viral Examples
Perplexity Computer: One Prompt, 19 Models, Entire Workflows Done 🖥️
Perplexity went quiet for two months. Now we know why. On February 25, CEO Aravind Srinivas launched Perplexity Computer — a multi-model agent orchestration platform that coordinates 19 different AI models to complete complex, long-running workflows entirely in the background.
How it works:
You describe the outcome — “build a marketing campaign,” “create a research report,” “develop an Android app.”
Computer breaks it into subtasks, spawns specialized sub-agents, and delegates each to the best-suited model.
The model roster: Claude Opus 4.6 (core reasoning), Gemini (deep research), Nano Banana (images), Veo 3.1 (video), Grok (speed tasks), ChatGPT 5.2 (long-context recall). Plus 13 more specialized models.
Runs for hours or months without re-prompting. Persistent memory. 400+ app integrations.
Sub-agents get a real browser, filesystem, and tool integrations. If one hits a wall, it creates more agents.
How it compares:
The Verge described it as existing “somewhere between OpenClaw and Claude Cowork.” Unlike OpenClaw (open-source, runs on your machine), Computer is fully cloud-managed — sandboxed execution, curated integrations, centralized safeguards. Unlike Claude Cowork (enterprise collaboration), Computer is consumer-facing and model-agnostic.
$200/month (Perplexity Max subscribers only). Pro and Enterprise rollout confirmed soon.
This is the first consumer product where you hand a single prompt to a system that delegates across every major AI lab’s best models. Whether it works reliably is TBD. But the architecture is genuinely novel. One early reviewer built 2 micro-apps, 4 research packets, and an automation workflow in a single night.
Source: VentureBeat | Fortune | Semafor | PYMNTS
Samsung Galaxy S26: The “Agentic AI Phone” Has Arrived 📱
Galaxy Unpacked 2026 (February 25, San Francisco) wasn’t just a phone launch — Samsung declared it “the beginning of truly agentic AI” on mobile. The Galaxy S26 series is Samsung’s third-generation AI phone, and this time the AI doesn’t just assist — it acts.
What’s new?
Now Nudge: Understands conversational context and proactively surfaces suggestions. Typing about dinner plans? It nudges you with restaurant bookings. No app switching required.
Gemini 3 agent tasks (beta): Offload multi-step tasks with a side-button press. In the live demo, Gemini read a group chat about pizza, built the order in a delivery app, and prepared the cart — all while the phone remained usable.
Three AI agents, one phone: Bixby (device agent with real-time web search), Gemini (agentic tasks), and Perplexity (”Hey Plex” wake word — first non-Google AI agent with wake-word access on Samsung).
World’s first mobile privacy display: Hardware-level screen that limits viewing angles on demand. Protect bank details, messages, passwords in public.
Performance leap: Snapdragon 8 Elite Gen 5 — 39% faster NPU, 19% faster CPU, 24% GPU boost. Up to 31 hours video playback. 75% charge in 30 minutes.
Pre-orders open now. General availability March 11, 2026. S26 starts at $900, S26+ at $1,100, S26 Ultra at $1,300.
The bigger picture
Samsung’s TM Roh said it best: “The technologies that change history fade into the background because they become infrastructure.” This is Samsung betting that AI on phones transitions from novelty feature to invisible utility. 350K+ peak concurrent viewers at the livestream suggest people agree.
Source: Samsung Newsroom | Google Blog | Engadget | TechRepublic
News of the week 🌍
Jack Dorsey Cuts 40% of Block, Blames AI 🪓 — The biggest AI-layoff story of the year just dropped. Jack Dorsey laid off 4,000+ employees (40% of Block’s workforce) saying “intelligence tools have changed what it means to build and run a company.” Block stock soared 20%+. Employees are furious — morale at “the worst in four years,” with workers calling top-down AI mandates “crazy.” A Harvard Business Review study cited in internal discussions found AI is actually intensifying workloads, not reducing them. This is the sharpest test yet of the “AI replaces headcount” thesis at scale
Meme or Not? A Toilet Company Is the Hottest AI Stock of 2026 🚽 — No, seriously. Japan’s TOTO — the company behind those legendary heated bidet seats — is up nearly 40% in 2026. Why? Their advanced ceramics division makes electrostatic chucks used in semiconductor fabrication for AI memory chips. Activist investor Palliser Capital called TOTO “the most undervalued and overlooked AI memory beneficiary,” claiming a $3.6B valuation gap. Ceramics now contribute ~50% of TOTO’s operating profit. When a toilet maker becomes an AI infrastructure play, you know the hype cycle has reached a new level.
Airbnb: 33% of Support Now Handled by AI 🏠 — During its Q4 earnings call, Airbnb revealed that a third of North American customer support is now managed by AI agents. Global rollout planned within a year, including AI voice agents. New CTO Ahmad Al-Dahle (ex-Meta Llama team lead) is steering the company toward becoming “AI-native.” Q4 revenue: $2.78B, up 12% YoY. CEO Brian Chesky: “AI is the best thing that ever happened to Airbnb.” Source
Microsoft Identifies “AI Memory Poisoning” Attack 🛡️ — Security researchers found that manipulated “Summarize with AI” links can embed hidden instructions that alter chatbot memory and bias future recommendations. 30+ organizations across finance, health, and SaaS sectors were affected. Microsoft implemented Copilot mitigations. Source
41% of Official MCP Servers Lack Authentication 🔒 — A security audit of 518 AI agent tools found that nearly half of MCP (Model Context Protocol) servers have no authentication. As agentic AI deployments multiply across enterprises, this is a ticking time bomb. Source
IBM Plans to Triple Entry-Level Hiring 📈 — Challenging the “AI kills junior jobs” narrative, IBM will triple entry-level hiring in 2026. Early-career employees will focus on client engagement and product development, augmented by AI, rather than repetitive coding. Source
OpenAI Drops gpt-realtime-1.5 🎤 — New real-time speech model with tighter instruction adherence for voice AI agents. If you’re building voice assistants, this matters. Source
Canva Acquires Cavalry and MangoAI 🎬 — Canva expands into 2D animation with two acquisitions, signaling a play for motion content creators. Source
Useful tools ⚒️
⭐ KiloClaw — Hosted OpenClaw. No Mac mini required. Cloud-based access to the viral open-source agent framework — one-click setup, no hardware needed. The “OpenClaw for everyone” moment.
Stitch by Google — Turn napkin sketches into production-ready UI in seconds. Google Labs’ new tool converts hand-drawn wireframes or sketches into working code instantly. Designers and devs, this one’s for you.
Notion Custom Agents — Anything you can do in Notion, your agent can now do for you — database queries, page creation, cross-workspace automation. The AI-powered workspace just got way more powerful.
Koidex — Know if a package, extension, or AI model is actually safe. Security scanning for the AI era. Essential for developers shipping AI-powered apps.
ChatPal — Practice speaking, get fluent! AI-powered language practice that feels like chatting with a patient tutor.
Weekly Guides 📕
Nano Banana 2: Official Blog & Getting Started — Google’s full breakdown of features, availability, and developer access
Build with Nano Banana 2: Developer Deep Dive — Pet Passport, Global Ad Localizer demos, API code samples, and partner integrations
Best Nano Banana 2 Prompts: eWeek’s Hands-On Guide — 6 tested prompts with real results, including multilingual text localization
Perplexity Computer: Hands-On Overnight Review — 2 micro-apps, 4 research packets, 1 automation — all built in one night
Samsung Galaxy S26: Agentic AI Features Deep Dive — Samsung Newsroom’s complete breakdown of Now Nudge, Gemini agents, and privacy display
AI Meme of the Week 🤡
related to the news mentioned earlier about TOTO & Ajinomoto
source: X
AI Tweet of the Week 🐦
(Bonus) Materials 🎁
Awesome Nano Banana Pro Prompt Collection on GitHub — Curated list of viral prompts for pet stickers, product posters, travel illustrations, and 6-panel comics
Google Revamped Flow Into a Full AI Creative Studio — Flow now combines image and video generation in one workspace
Nano Banana 2 Viral Examples Roundup by BGR — 10 wildest early outputs with the tweets that posted them
If you missed our previous updates, don’t worry, here they are:
Sonnet 4.6, Grok’s 4 Agents, and Meta’s Manus in Ads Manager | Weekly Digest