分类筛选
找到 1116 个 Demo
Google AI Studio just became a real app builder. �...
Google AI Studio just became a real app builder. 🧵 Here’s what changed: 1. AntiGravity coding agent added 2. Firebase backend auto-setup 3. Authentication built-in 4. API integrations supported 5. Real-time collaboration enabled Prompt → product That’s the workflow now. Save this video, you’ll stay ahead. Want the SOP? DM me. 💬
ElevenLabs Flow: first I generate two character sh...
ElevenLabs Flow: first I generate two character sheets in Nano Banana‑2 and Seedream‑4.5 to compare and pick the best heroine, then a track in Eleven Music, Gemini 2.5 to plan the scenes + close‑ups, and finally video in Kling 3.0/Omni 3. https://t.co/Vgo8UiUrHZ @ElevenLabs @ElevenCreative
Turn your PDFs into a podcast 📄🔈 We built a sim...
Turn your PDFs into a podcast 📄🔈 We built a simple tool that integrates Gemini 3.1 voice with LlamaParse document parsing. Transcribe, read, and interact with all of your documents through voice! Repo: https://t.co/MFKol8W79w Sign up to LlamaParse: https://t.co/e0hRfFjbqs https://t.co/cCd1XYorWF
🏆 2nd place at the Google Hackathon. In 24 hour...
🏆 2nd place at the @DeepStationAI Google Hackathon. In 24 hours I built a free tool for Miami residents to find housing, food, healthcare, jobs & education in their language, in minutes. Built with: → @Google ADK → Gemini 2.5 Flash → 5 specialized AI agents → Federal, Florida & @MiamiDadeCounty programs → English, Spanish & Creole
OTD 3/23/1965: launched #Gemini3, the first crewe...
OTD 3/23/1965: @NASA launched #Gemini3, the first crewed #Gemini mission, which completed the first crewed orbital maneuvers in #space. https://t.co/BoydzUo5vT #spaceflight #spaceexploration #exploration https://t.co/iVkcscTDok
100+ AI Tools to replace your tedious work: 1. Re...
100+ AI Tools to replace your tedious work: 1. Research - @ChatGPTapp - YouChat - @abacusai - @perplexity_ai - Copilot - Gemini 2. Image - @higgsfield_ai Soul - GPT-4o - Midjourney - Grok 3. Productivity - @GammaApp - Grok 3 - Perplexity AI - Gemini 2.5 Flash 4. Writing - Jasper - Jenny AI - Textblaze - Quillbot 5. Video - Klap - Kling - @invideoOfficial - HeyGen - Runway 6. Meeting - Tldv - Otter - Noty AI - Fireflies 7. SEO - VidIQ - Seona AI - BlogSEO - Keywrds ai - Outrank AI 8. Presentation - @decktopus - Slides AI - Gamma AI - Designs AI - Beautiful AI 9. Design - @canva - Flair AI - Designify - Clipdrop - Autodraw - Magician design 10. Audio - Lovo ai - @elevenlabs - Songburst AI - Adobe Podcast 11. Marketing - Pencil - Ai-Ads - AdCopy - Simplified - AdCreative 12. Startup - Tome - Ideas AI - Namelix - Pitchgrade - Validator AI 13. Social media management - Tapilo - Typefully - Hypefury Follow @rosemoni18 👤 for more such amazing stuff ❤️
Google just launched Gemini 3.1 Flash Live, a new...
Google just launched Gemini 3.1 Flash Live, a new realtime model built for voice and vision agents. After a year of work on model quality, infrastructure, and UX, they're calling it a step-function improvement in quality, reliability, and latency. The race to build the best real-time AI agents just got more intense!
I built the entire UI for my SaaS in 30 minutes....
I built the entire UI for my SaaS in 30 minutes. Day 5 of building ThinkBoard in public. Once my design system was locked in, I switched to Gemini 3.1 Pro (cheaper and fast), gave it the UI development plan + some design inspiration screenshots, and Cursor shipped the entire thing in 15 minutes straight. I walked through the user journey, fixed a few small issues, and BOOM. Done. This is the speed you need in 2026. Stop spending weeks designing UI for an idea nobody wants yet. Ship fast, validate, THEN hire designers and spend thousands. If no one cares about your product, every extra week you spend is just burning runway and killing momentum.
🚀 The team just added Gemini 3.1 to the Live API...
🚀 The @GoogleDeepMind team just added Gemini 3.1 to the Live API, so we built a small demo showing how Gemini voice agents can plug directly into the document processing ecosystem powered by LlamaIndex. 🔥 In this example, we integrate LiteParse to enable fast, fully-local document parsing. With our TUI-based voice assistant, you can literally talk to your terminal: - Speak commands - Trigger live document parsing via tool calls - Hear the agent read back results in real time 🔊 The assistant can extract content from single files or entire folders, leveraging the lightning-fast local parsing that LiteParse provides ⚡ Take a look at the demo👇 👩💻 GitHub repo https://t.co/ySmenP2HoY 📚 LiteParse docs https://t.co/NlpoI4CqEq
🚨 HOLY SHIT... Google AI just proved that fine-tu...
🚨 HOLY SHIT... Google AI just proved that fine-tuning Gemini 2.5 made it dumber on hard queries. > Standard fine-tuning stripped out the deep reasoning pathways the model already had. Replaced them with shallow pattern matching. The fine-tuned version scored lower than the base model it started from. > Meanwhile a 7B model trained on Chain-of-Thought data nearly closed the gap entirely. > The experiment was simple. Google's AI team fine-tuned Gemini 2.5 Flash on a standard text-to-SQL dataset and tested it on 600 queries weighted toward complex joins, nested subqueries, and multi-step logic. The base model scored 73.17%. The fine-tuned version scored 72.50%. Basically unchanged on the full set. But on the hardest 40 queries, the fine-tuned model dropped from 62.5% to 57.5%. The model got measurably worse at the exact tasks fine-tuning was supposed to improve. > The explanation matters. Large models already have deep, multi-step reasoning baked in from pretraining. When you fine-tune them on standard input-output pairs without explicit reasoning traces, you're not adding capability. You're forcing the model to abandon its reasoning pathways in favor of memorizing structural templates. On easy queries, the template is good enough. On hard queries, the model tries to recall a pattern instead of reasoning through the problem. That's the failure mode. Google calls it "representation collapse." > The small model result is the real finding. Qwen 7B started at 36.17% accuracy. Standard fine-tuning pushed it to 45.33%. Still not production-grade. Then they added Chain-of-Thought reasoning steps to the training data: query analysis, table selection, column matching, join logic, self-validation. The same model hit 54.5%. An 18-point absolute improvement over baseline. The gap between a 7B model and Gemini 2.5 Flash went from 37 points to 19 points just by teaching the small model how to think through the problem instead of memorizing the answer. The takeaway for anyone building on top of models: → Fine-tuning large models on task data without reasoning traces can hurt them on hard cases → Fine-tuning small models on reasoning traces is how you extract real capability → The data format matters more than the dataset size → CoT training isn't just a prompting trick it's a training strategy Large models don't need more examples. They need their reasoning preserved. Small models don't need more data. They need to be shown how to think.
100+ AI Tools to replace your tedious work: 1. Re...
100+ AI Tools to replace your tedious work: 1. Research - ChatGPT - YouChat - Abacus - Perplexity - Copilot - Gemini 2. Image - Higgsfield AI Soul - GPT-4o - Midjourney - Grok 3. Productivity - Gamma - Grok 3 - Perplexity AI - Gemini 2.5 Flash 4. Writing - Jasper - Jenny AI - Textblaze - Quillbot 5. Video - Klap - Kling - InVideo - HeyGen - Runway 6. Meeting - Tldv - Otter - Noty AI - Fireflies 7. SEO - VidIQ - Seona AI - BlogSEO - Keywrds ai - Outrank AI 8. Presentation - Decktopus - Slides AI - Gamma AI - Designs AI - Beautiful AI 9. Design - Canva - Flair AI - Designify - Clipdrop - Autodraw - Magician design 10. Audio - Lovo ai - Eleven labs - Songburst AI - Adobe Podcast 11. Marketing - Pencil - Ai-Ads - AdCopy - Simplified - AdCreative 12. Startup - Tome - Ideas AI - Namelix - Pitchgrade - Validator AI 13. Social media management - Tapilo - Typefully - Hypefury - TweetHunter Follow @RodmanAI for more such amazing stuff ♥️
🚨 BREAKING: Someone built a coding editor that le...
🚨 BREAKING: Someone built a coding editor that lets you use Claude Opus 4.6, GPT-5.4 and Gemini 3.1 Pro without paying a single dollar in API fees. It's called Glass and it just made Cursor, Windsurf, and every AI IDE look like a rip-off. Here's how it works (in plain English):↓