← Back to leaderboard
shruti_0810

Shruti Codes

Verifiedactive
@shruti_0810

Regular Tech & AI Updates 💻 | Sharing Free Resources 🌐 | Collabs: tripathishruti810@gmail.com | Daily Tech news

RosterAI Agents & Productivity ToolsTech Career & Coding Education
105,798 followers · 401 following · Last polled 2h ago
Perf. Score
45.8
Rank #264
Value Score
80
#55 of 273 priced
Median views
5.2K
p25 2.2K · trailing 7d
Followers
106K
-0.2% 7d
ER (impr.)
0.44%
eng ÷ impressions
ER (foll.)
0.09%
eng ÷ followers
Est. CPM
$4.81
QT-rate basis
WoW views
-23.8%
▼ falling

Rates & value · est. from 5.2K median organic views

underpriced
Quote tweetbasis
$25
$4.81 / 1K views
Post
$35
$6.73 / 1K views
Thread
$55
$10.58 / 1K views
Retweet
—
no CPM
Value Score 80 · rank #55 of 273 priced$312.5 / 1K engagementsrate age unknown (pre-audit import)

Trends

Follower growth
Median views / day
Engagement rate
View distribution · trailing-7d posts

Each blue dot is one organic post. The Performance Score ranks by the median (5.2K), not the mean (21K) — so a single viral post doesn't inflate a creator's reach. The p25 floor (2.2K) is the level 75% of posts clear.

Recent posts · ranked by views

PostPostedViewsLikesRepostsRepliesQuotesBookmarksER
I’ve been catching up on posts from Apsara Conference 2026, and one thing really stood out to me: AI is starting to feel less like something that lives on a screen. When people say “embodied AI,” I usually think of humanoid robots from companies like Unitree or Tesla. But Apsara showed a much wider picture. AI is showing up in everyday products and real-world applications, alongside the chips, compute, cloud infrastructure and AI agents making all of it possible. My friend who attended the conference was telling me how much was happening across the floor, and honestly, Shenzhen and Yiwu seem to have sprouted wings. What I found most interesting wasn’t just what AI can do today. It’s seeing where it’s starting to show up next.frozen29d ago281K5519193140.04%
Everyone talks about smarter robots. But here’s the real challenge: How do you teach a robot to perform human tasks in the real world? 🤖 The answer starts with data. That’s what makes @Maniformer work interesting. They’re building a one-stop data infrastructure for Physical AI, bringing together 3 key sources: → Real-world data: How robots perceive and interact with their environment → Simulation data: Scalable synthetic data to train and improve models → Human demonstration data: Real human movements that help robots learn complex skills But what really caught my attention is their MEgoVista research It reconstructs 3D hand and head movements from first-person recordings, reporting 4.29 mm finger articulation error against motion-capture ground truth. Why does this matter? Because a robot doesn’t just need to recognize what a human is doing. It needs to understand the movement well enough to learn how to do it itself. And scaling that process could be a major step toward more capable robots. Better data → Better learning → More capable robots. Physical AI is more than just building better hardware. The data behind it matters just as much. Explore MEgoVista research: https://t.co/D45XgcWVhU
frozen
10d ago
214K
71
17
17
4
13
0.06%
When your friend finds you a $0.29 AI video deal 😂 Ready to bring your next video idea to life? Try Seedance directly on Dreamina. Use my official creator partner link to access the $0.29 offer 👇 https://t.co/fbcNahDZok #DreaminaOnly29Cents #Dreamina #DreaminaPartner #Seedance1d ago140K217143626170.22%
A video can need just a few finishing touches and still leave you with plenty to do. With CapCut Web × Codex, I can describe those changes in everyday language. Ask for captions, switch the filter, add transitions, and have Codex call CapCut’s libraries for music, sound effects, and visual effects. ChatGPT helps interpret the request. CapCut provides the resources and editing tools. It’s a more straightforward way to move through the final adjustments and cut down the time spent on routine editing. #capcutweb #capcutchatgptplugin #capcutcodex #capcutmarketing2d ago121K5723121130.09%
What if the biggest shift in SaaS is that humans stop operating the software? 👀 https://t.co/DgOEvIJd7v’s launch film explores an interesting idea: a future where AI agents don't just use business software, but help build it and execute workflows. That's the promise behind Headless SaaS. With Proto, https://t.co/DgOEvIJd7v is working toward a model where you describe what your business needs, and an agent helps create the app and get the work done. If this vision becomes reality, it could change how businesses build, buy, and use software. Are we moving from software built for humans to software operated by AI agents? Explore https://t.co/DgOEvIJd7v: https://t.co/M0BdbdHUQefrozen7d ago80K71103180.12%
What happens when you clean up voice before sending it to speech-to-text? Krisp just released an open benchmark testing this across: • 265 real recordings • 11 STT configurations • Call centers, offices & moving cars The result? 📉 ~73% lower word error rate across the benchmark. And the full dataset is now public on Hugging Face. This is especially interesting for voice agents dealing with real-world noise: 🎙️ Background conversations ⌨️ Keyboards 🚗 Traffic ❄️ AC noise 🗣️ People speaking nearby Because in real-world environments, speech-to-text doesn't always get clean audio. See the benchmark + results: 🔗 https://t.co/oJfQ012rKefrozen28d ago43K7268560.23%
Claude Code can build your entire app. Then you open the cloud console and suddenly it’s 2018 again. 💀 Provision servers. Configure infra. Monitor traffic. Scale manually. InstaCloud is going after exactly that gap: let the agent deploy and manage the infrastructure too. This feels like an important piece of the agentic coding stack.frozen23d ago40K772316280.31%
Akshay Kumar after realizing he can finally change the expression without reshooting the scene 💀 DRAMA: bro, we got you. https://t.co/v5lzDe13gjfrozen15d ago36K67169340.28%
10 OPEN-SOURCE PROJECTS MAKING AI MORE USEFUL EVERY DAY. Not another list of chatbots. These projects are making AI work with files, browsers, data, memory and real workflows. 1. GenOffice An AI-native office suite with Docs, Sheets, Slides, PDF and Markdown, plus an AI agent and CLI. https://t.co/CAsxClzbkX 2. OpenMAIC Turn a topic or document into an interactive multi-agent classroom with AI teachers, classmates and multimedia. https://t.co/0jDKEFk9o8 3. WeKnora Build intelligent knowledge applications using document understanding, retrieval and agentic workflows. https://t.co/e04i2RGBq4 4. Graphiti A temporal knowledge graph built for AI agents to understand entities, relationships and changing information. https://t.co/bufepyERzu 5. Cognee Add a deeper memory layer to AI applications using knowledge graphs and connected data. https://t.co/wFeWvwYoJr 6. Browser Use Let AI agents navigate websites, interact with pages and complete browser-based tasks. https://t.co/bjo4KSoOpL 7. Open WebUI A self-hosted AI workspace that connects models, tools, knowledge bases and agent workflows. https://t.co/iWj9O37fEY 8. PageIndex A reasoning-based approach to navigating complex documents using structured tree-based retrieval. https://t.co/4qaEyffsey 9. Agent-Reach Give AI agents better access to information from the web and other sources. https://t.co/bYkBfERYlU 10. Qwen Audio Agent A real-time voice runtime that pushes AI agents toward listening, speaking and real-time interaction. https://t.co/G0r2T161Vc The shift is happening: AI that answers → AI that works. And open source is making that transition easier for everyone to experiment with. Bookmark these. There are some serious gems here.frozen26d ago34K421762228684.05%
* ₹2,500 crore worth of shares reserved for eligible Reliance Industries shareholders. * Jio offers multiple digital services beyond mobile and internet. * 100% fresh issue of up to 27 crore shares — no Offer for Sale (OFS). * IPO proceeds will go to the company, primarily to repay debt. 📅 October 21–23, 2026 | 💰 Price band: ₹1,065–₹1,119 🔗 More details: https://t.co/7fG1r034e0frozen7d ago32K8279030.32%
The future of AI coding might not live in the cloud. It might live on your laptop. Recursive just released Flint, a 40B coding model built to run locally at high speeds. 125+ tokens/sec on a MacBook and 150+ on an RTX Spark at short contexts. Your code stays on your hardware. No per-token API bill. The wildest part? Recursive's Eureka Machine helped build Flint in under four months. AI helping build better AI. We're entering a new era.1d ago24K1705390.14%
What happens when AI stops just giving medical answers and starts handling the workflow? That’s what makes this launch interesting. Nolla is starting with acne care in Utah, bringing together intake, assessment, treatment planning, prescribing when appropriate, and follow-up. This could be a meaningful step for AI in healthcare.frozen17d ago23K9416131120.60%
This might be one of the most useful OpenAI interviews I’ve seen lately. 50+ minutes covering ChatGPT, Codex, agents, AGI, AI safety, the future of work, and what OpenAI is building next. The “what happens when AI starts building AI?” part is especially interesting. Worth watching the full thing.frozen21d ago20K762483210.65%
AI won’t replace designers. It will change how they create. MagicOS takes a different approach by making AI a collaborator, not the one in control. A shared canvas, real-time voice interaction, and your entire design process in one place. This is a much more natural way to work with AI. Excited to see where this goes!frozen9d ago20K42166160.36%
AI just crossed an uncomfortable line. @tavus gave me early access to Griffin, and this isn’t your usual chatbot. It can see you. React to you. Interrupt you. Even understand your gestures in real time. 45% of people thought Griffin was a real person in a Turing test. This is going to change how we think about AI conversations.frozen21d ago20K751912370.59%
I DON'T UNDERSTAND WHY PEOPLE ARE NOT USING JEV SKILLS. 20 must-use JEV skills for your agent setup: 1. jev-ultrafast → browser agent https://t.co/MBcBTKeFy4 2. fast-jev-compaction → compress context https://t.co/zyoPJwFm4I 3. json-render → generative UI https://t.co/WZGubwlfHC 4. typesafe-mcp → add JEV to any MCP client https://t.co/2W5nTIJoKB 5. jev-mcp → judgment tools for agents https://t.co/YXhPx4wxo0 6. semdecide → CLI classifier https://t.co/HKJWa1vwRc 7. jev-codex-router → route coding tasks to the right model https://t.co/44SyIv5SGK 8. winnow → remove useless context https://t.co/p3FfLYWWjP 9. jev-review → sort code review issues https://t.co/sfiFeIb3jN 10. blink → find your way around a repo https://t.co/sFccz5ic5v 11. agent-desktop → desktop automation https://t.co/P0MBHM9Rrd 12. typesafe-mario → agent that plays Super Mario https://t.co/YR5VQg4zLk 13. jev-drone → control a drone https://t.co/kxa2eB7uPZ 14. onevonejev → browser FPS game https://t.co/vZIk02EBcZ 15. jev-trader → market-making agent https://t.co/FtV3EJHGti 16. prism → spot liquidity signals https://t.co/f7rxDIabzd 17. neo4jev → move through knowledge graphs https://t.co/wezQktIcds 18. jev-curate → filter training data https://t.co/yuJyoBCed8 19. canny → check if an agent really finished the task https://t.co/Gc0tq10h1x 20. killmyidea → score your startup idea https://t.co/PAk0GhpFxV Bookmark this list if you're building with JEV.frozen29d ago18K209291913283.28%
10 GITHUB REPOS THAT TURN 2D IMAGES INTO 3D ASSETS Imagine uploading one image and getting a 3D model you can use in your creative projects. These 10 open-source projects are making image-to-3D generation more accessible: 1. TRELLIS Generate detailed 3D assets from images, including textured meshes and 3D representations. https://t.co/3iVim8RZpQ 2. Hunyuan3D-2 Create 3D shapes and textures from images or text prompts. https://t.co/1fJO8GvAkl 3. TripoSR Reconstruct 3D models from a single image in seconds on suitable hardware. https://t.co/C7dkcECLfs 4. Stable Fast 3D Turn object images into textured 3D meshes quickly. https://t.co/UgU1Icb8zt 5. InstantMesh Generate 3D meshes from a single image using multi-view reconstruction. https://t.co/uRPCf1FayU 6. Wonder3D Generate consistent views and reconstruct textured 3D objects. https://t.co/kI0I1V7sFz 7. CRM Explore single-image 3D reconstruction with a creative reconstruction pipeline. https://t.co/cBBFS3WnPZ 8. Unique3D Transform 2D images into detailed 3D assets with generated geometry and textures. https://t.co/4RryTbOzbq 9. OpenLRM Reconstruct 3D representations from images using large reconstruction models. https://t.co/NTqS0voXNh 10. TRELLIS Explore rich 3D generation with outputs such as textured meshes and 3D Gaussians. https://t.co/3iVim8RZpQ From game assets to digital art, these tools open up exciting creative possibilities. Save this list. Your next 3D project might start with just one image.frozen11d ago14K21731703113.98%
AI has learned to read, write, and code. But understanding what’s actually happening in a video? That’s a different challenge. A model can describe a scene and still miss the logic behind it. @ElorianAI wants to close that gap with Glimpse, its first visual thinking model designed to reason over images and videos. Incredibly interesting to see where the industry is heading.frozen6d ago14K51116030.50%
the gap between “AI can help you write an email” and “AI is autonomously testing your mobile app on a foldable phone in both screen modes” happened in like 2 years @autosanaai understood the assignment. first platform to support iPhone Duo testing. this is what agentic AI actually looks like when its not just a buzzwordfrozen16d ago14K7021103110.85%
10 GITHUB REPOS THAT CAN REPLACE EXPENSIVE SOFTWARE You might be paying for tools that already have open-source alternatives. Here are 10 worth checking out: 1. AppFlowy Notion alternative for docs, wikis, databases, boards and more. https://t.co/w6XwlRjTD7 2. NocoDB Turn databases into a spreadsheet-style workspace like Airtable. https://t.co/AP3bwDB1L2 3. Listmonk Self-hosted newsletter and mailing-list management. https://t.co/LrYD177mXT 4. https://t.co/Qgx4dAbfC0 Open-source scheduling and booking platform. https://t.co/VRf7TNFdXM 5. Formbricks Build surveys, forms and in-app feedback without relying on a hosted platform. https://t.co/TX4tX4nSiu 6. Activepieces Open-source automation platform for connecting apps and building workflows. https://t.co/YxDreo6lyf 7. Outline Clean, self-hosted knowledge base for teams and documentation. https://t.co/3xTnl0dBJ1 8. Plane Open-source project management for issues, sprints, cycles and roadmaps. https://t.co/efIaTMz03f 9. Documenso Self-hosted document signing and e-signature platform. https://t.co/pQ0TvUXLXw 10. Chatwoot Open-source customer support with live chat, shared inbox and help center. https://t.co/S2CavYmVvS The interesting part? You don't always need another expensive SaaS subscription. Sometimes, the open-source alternative is already on GitHub. Save this list before your next software renewal.frozen20d ago13K16537702613.62%
We might be approaching a world where apps become optional. Imagine describing what you want, and AI builds a working app around your needs instead of making you adapt to someone else's software. No endless downloads. No juggling 20 different tools. Just tell it what you need and get to work. Wabi is an interesting step in that direction 👀frozen3d ago13K5730020.48%
Your AI agent shouldn’t be trapped in one app. 👀 Imagine an agent that can call you, text you, email you, and reach you on WhatsApp or Telegram. No app switching. No platform limitations. Just agents that meet you wherever you are. Photon 2.0 is making that future feel a lot closer. And $100 in free credits for beta users? That’s a pretty good time to experiment. 🔥frozen7d ago13K320020.05%
10 GITHUB REPOS THAT MAKE DIGITAL HUMANS LOOK REAL 🤯 These open-source projects let you experiment with talking avatars, lip-sync, facial animation, and AI-generated portrait videos. 1. LiveTalking Build interactive AI avatars that respond and speak in real time. https://t.co/bkqVh7swMW 2. MuseTalk Generate realistic lip-sync by matching mouth movements to audio. https://t.co/9ers9Frr6G 3. Hallo Turn a portrait into a talking video using audio-driven facial animation. https://t.co/0nzsxG533S 4. SadTalker Animate a still image with speech, facial expressions, and head movements. https://t.co/bh0vq3L4FO 5. LivePortrait Bring portraits to life with expressive facial movements and motion references. https://t.co/wPWpkTo3Da 6. Wav2Lip Synchronize lip movements with spoken audio to create convincing talking-face videos. https://t.co/pLsMQYoFHZ 7. EchoMimic Generate expressive talking portraits using audio and visual references. https://t.co/GpkDOqHxO7 8. AniPortrait Create animated human portraits and virtual characters driven by audio. https://t.co/u4Vcf9M9Cb 9. Hallo3 Explore long-form, audio-driven portrait video generation with expressive motion. https://t.co/bbomkAF9gh 10. Hallo2 Generate animated talking portraits using audio and diffusion-based video generation. https://t.co/tsk8v9RdRM Think about the possibilities for virtual presenters, AI characters, interactive assistants, and video creation. The technology is moving fast, and these repos give developers a way to experiment with it themselves. Bookmark this list. Which project would you try first?frozen9d ago13K18745912814.13%
We’ve spent years teaching AI to answer questions. The next big leap? Teaching AI to work as a team. Think about it. Real business operations involve people, tools, approvals, and dozens of moving parts. One AI assistant can only go so far. Now imagine specialized AI agents working alongside your team, sharing context, handling workflows, and getting things done in the same chat. That’s the shift @zenzap_chat is betting on. AI isn’t just another tool for your team. It could become part of the team itself.frozen8d ago13K1781010.22%
An API error shouldn’t mean an AI agent has to call a human for help. The problem is, agents often get an error without enough context to know what went wrong or how to fix it. Restless tackles this by adding recovery steps and relevant docs directly to failed API responses. Small change in the error experience, big difference in how agents can work. @restless_hqfrozen10d ago13K531271100.66%
The hardest part of a trip shouldn't be planning it. 🌍 Flights, restaurant bookings, travel points, price drops, cancellations… Odessia helps handle the details so you can spend less time organizing and more time looking forward to the trip. Explore @OdessiaTravel at https://t.co/lOvxGXYGDtfrozen8d ago12K302000.04%
10 GITHUB REPOS THAT REPLACE TOOLS YOU'RE PROBABLY PAYING FOR Some tools cost $10–$100+ every month. Their open-source alternatives are sitting on GitHub. 1. yt-dlp Download videos from supported sites for personal/offline use. https://t.co/GvxDZnCvlm 2. Ollama Run AI models locally on your own machine. https://t.co/JO8NMgKVp1 3. Fooocus A simple open-source interface for AI image generation. https://t.co/9e5SaYeCoh 4. Whisper Open-source speech recognition that can transcribe audio locally. https://t.co/vQ0brcdBNz 5. Plausible Privacy-focused, self-hostable web analytics. https://t.co/1qOQLYfQaJ 6. AppFlowy Open-source workspace for notes, docs, databases and productivity. https://t.co/w6XwlRjTD7 7. Penpot Open-source design and prototyping platform for collaborative teams. https://t.co/bwQxBLNiC8 8. n8n Build powerful workflows and automate tasks across different services. https://t.co/wHYb9xgDa6 9. https://t.co/Qgx4dAbfC0 Open-source scheduling platform for booking and appointments. https://t.co/oq5mqNXrt4 10. Bitwarden Open-source password manager with a strong free offering. https://t.co/BNwD94Bb9b Save this list.frozen18d ago12K91241311242.10%
The next generation of AI may need to understand physics, not just language. Odyssey-3 is tackling that problem with a world model that predicts how environments evolve as actions happen in real time. Odyssey-3 Pro reported a 66.1 Physics-IQ Verified video-to-video score, the highest reported score. Try it here: https://t.co/z1FTctz1lSfrozen14d ago12K2481020.30%
This team literally built a QA layer for AI voice. @OnepinAI checks every line for naturalness, clarity and pronunciation, fixes mispronounced words directly, and routes across 30+ tested models. This is how it works from a pasted script 👇 https://t.co/oU8gjqGK1ffrozen21d ago12K2337390.38%
AI agents having a phone book is a pretty wild concept. Instead of every agent being isolated, they can discover other agents, connect with them, and actually get things done together. This feels like an important step toward an agent-to-agent internet.frozen14d ago11K82143030.90%
Everyone wants a smarter AI model. But what if the real bottleneck is the computer it's using? AI can reason through complex problems, yet executing real work still means screenshots, clicks, and endless verification loops. @PineAIAssistant is taking a different approach: rebuilding the computer for AI, not rebuilding the prompt. The next AI breakthrough might not be a smarter model. It might be giving AI a computer actually built for it.frozen13d ago11K662012010.89%
Not every brand commercial needs to feel like an ad. This feels like a proper piece of culture. 🔥frozen23d ago11K421514010.65%
that little pinky promise at the end was such a good touch. it’s funny how something so simple can represent an agreement people actually take seriously. makes the whole “agreements make history” line hit differently. #adfrozen23d ago11K6841000.67%
5 GITHUB REPOS THAT MAKE DATA ANALYSIS WITH AI MUCH EASIER You don't need a huge data stack to start analyzing data with AI. These 5 open-source projects are a great place to begin: 1. PandasAI Ask questions about your datasets using natural language and generate analysis with Pandas. https://t.co/NfLzEJnpzY 2. Data Formulator Turn your data into charts and visualizations using natural language. https://t.co/UG4wzPqo8E 3. Open Interpreter Let AI interact with your computer and execute code to work with data. https://t.co/JrkV9kbEBs 4. MindsDB Connect AI with databases and build intelligent data workflows. https://t.co/7RSUz4xnXD 5. Vanna Use natural language to generate SQL and query your databases. https://t.co/sMp0kLI0Mj Data → AI → Insights Save these if you're building AI-powered data workflows.frozen19d ago8.8K123271721443.55%
10 OPEN-SOURCE PROJECTS MAKING AI MORE USEFUL EVERY DAY 👇 AI is moving beyond answering questions. These GitHub repos help it work with documents, browsers, knowledge, memory, and real-world tasks. 1. GenOffice AI-powered workspace for Docs, Sheets, Slides, PDFs, and more. https://t.co/CAsxClzbkX 2. OpenMAIC Turn any topic or document into an interactive AI classroom. https://t.co/0jDKEFk9o8 3. WeKnora Build intelligent knowledge apps with document understanding and AI workflows. https://t.co/e04i2RGBq4 4. Graphiti Build temporal knowledge graphs that help AI agents remember changing information. https://t.co/bufepyERzu 5. Cognee Give AI systems a deeper memory layer using connected data and knowledge graphs. https://t.co/wFeWvwYoJr 6. Browser Use Let AI agents navigate websites and complete browser tasks. https://t.co/bjo4KSoOpL 7. Open WebUI A self-hosted AI workspace for models, tools, documents, and agent workflows. https://t.co/iWj9O37fEY 8. PageIndex Navigate complex documents with reasoning-based, tree-structured retrieval. https://t.co/4qaEyffsey 9. Agent-Reach Help AI agents access information across the web and other sources. https://t.co/bYkBfERYlU 10. Qwen Audio Agent Explore real-time AI voice interactions with listening and speaking capabilities. https://t.co/MkHnFu0JyT Bookmark these repos. Which one are you exploring first?frozen11d ago8.3K862680942.56%
AI has transformed how we work. Now, what if it could change how much we pay for everyday essentials? Aonic Club is taking a different approach: use AI to cut out the layers between manufacturing and your doorstep. 30 electrolyte sticks: • Liquid I.V. on Amazon: $46.86 • Aonic Race: $9.35 The goal? Products at cost. No traditional retail markup. Protein and supplements are just the beginning. @maxjmeier @AonicClub END THE MARKUP.frozen9d ago7.8K611311171.20%
Job hunting is one of those things that desperately needed better automation. An AI recruiter that works 24/7, finds relevant roles, prepares applications, and even helps with interviews is a pretty compelling use of AI. Sage looks interesting.frozen17d ago7.5K231082130.75%
AI infrastructure is scaling fast, and that means more compute for builders everywhere. More GPU capacity across key markets = more room to build, experiment, and ship AI applications. Big milestone for @gmi_cloud with a $668M Series B.frozen22d ago6.4K932812172.22%
Cold outreach is still way too manual. Runable is turning it into something much more automated from calls and emails to finding the right local numbers. Less busywork. More conversations. This is where AI-powered outreach gets interesting.frozen21d ago6.2K682021361.90%
5 GITHUB REPOS TO BUILD YOUR OWN LOCAL AI APPS Want to run AI on your own machine instead of relying entirely on cloud services? These 5 open-source projects are a great place to start: 1. Ollama ↳ https://t.co/pCQefcLv45 Run large language models locally with a simple setup. Great for experimenting with models and building AI-powered applications. 2. Open WebUI ↳ https://t.co/iWj9O36HPq A self-hosted AI interface that lets you chat with local models and connect different AI backends. 3. llama.cpp ↳ https://t.co/tHe3u7Hize Run LLMs efficiently across different hardware, including machines without high-end GPUs. 4. LocalAI ↳ https://t.co/4OpYFcpmeJ Build self-hosted AI services with API compatibility for popular AI tools and support for multiple model types. 5. Jan ↳ https://t.co/TQNdy2VyJD A desktop AI assistant that runs models locally and helps you explore private, offline AI workflows. Local models. More control. Fewer cloud dependencies. Save this list if you're building your own AI stack.frozen12d ago6.2K781780662.73%
HOLY SHIT. WHAT IF AI COULD HELP BUILD BETTER VERSIONS OF ITSELF? These 10 GitHub repos explore coding agents, automated experiments, evolutionary search, and AI systems that improve through feedback. 1. SWE-agent Give AI a GitHub issue and let it navigate code, edit files, and attempt a fix. https://t.co/XPI6I5IyXH 2. Aider Let AI modify code across your repository with Git integration. https://t.co/flZ0KbHYaj 3. OpenHands Build AI coding agents that use terminals, edit files, and run commands. https://t.co/0R04hvIEgO 4. DSPy Optimize AI pipelines through systematic experimentation instead of manual prompt tweaking. https://t.co/XQvfkWCSJD 5. TextGrad Improve prompts and multi-step AI workflows using feedback-inspired optimization. https://t.co/aoOEZWfHHv 6. EvoAgentX Build, evaluate, and optimize multi-agent workflows. https://t.co/hx3Wj9uFQH 7. ADAS Explore how AI can discover better agent designs automatically. https://t.co/hnEnsAwwY8 8. OpenEvolve Generate, evaluate, and refine code using evolutionary search. https://t.co/YtmtlmBHcB 9. SWE-smith Generate software-engineering tasks and datasets for training coding agents. https://t.co/UcbzsUtuJb 10. AutoGPT Experiment with autonomous agents that plan tasks, use tools, and execute multi-step workflows. https://t.co/fY16qO0Yl0 The shift is fascinating: Bookmark these repos. Which one would you experiment with first?3d ago6.0K642780752.92%
The part i care about isnt that it runs on a schedule.. its that i dont have to come back and finish the task myself. the email gets answered. the follow-up gets sent. the meeting gets booked. thats the difference between a reminder and actually getting the work done. https://t.co/c0xXtCzi9kfrozen28d ago5.6K3693381.06%
10 GITHUB REPOS FOR BUILDING YOUR OWN AI INFERENCE STACK All open-source. All worth bookmarking. If you're building AI agents, local LLM apps, or production inference infrastructure, these are worth knowing: 1. SIE — One inference layer for multiple AI models and agent workloads. https://t.co/LjOKuM9dtm 2. vLLM — High-throughput LLM serving with batching, quantization and distributed inference. https://t.co/9kQt9PxS3b 3. SGLang — Fast serving for LLMs and multimodal models. https://t.co/npVOHoJenl 4. llama.cpp — Run LLMs locally across laptops, servers and edge hardware. https://t.co/tHe3u7HQoM 5. Ollama — Simple way to run and interact with local models. https://t.co/pCQefcM2TD 6. LocalAI — Open-source API for LLMs, vision, audio and other AI workloads. https://t.co/4OpYFcpU4h 7. Triton Inference Server — Production model serving across multiple frameworks. https://t.co/q19E2A5huC 8. BentoML — Build, package and deploy production inference services. https://t.co/Ngpz6Pqse0 9. KServe — Kubernetes-native platform for AI inference. https://t.co/2elvxey1hC 10. SIE — Worth another look for its broader inference capabilities across agent workloads. https://t.co/LjOKuM9dtm Bookmark these for when you build your own inference stack.frozen28d ago5.4K8419241853.93%
What if your product showed up in the videos everyone shares? 93 Halloween viral presets. Add your logo and it's in there. Made to get views. Made to sell. Skip the costume and the camera. Don't miss 50% off BuzzHooks.7h ago5.4K1233010.35%
THESE 10 GITHUB REPOS CAN TURN AI INTO A FULL-TIME RESEARCHER. Imagine an AI that searches the web, reads documents, finds relevant sources, and turns everything into useful answers. You can start building it with these 10 open-source repos: 1. GPT Researcher Generate detailed research reports https://t.co/3unuxPrAkt 2. Tavily MCP Give AI agents web search tools https://t.co/c5DRVIFTyO 3. Oxylabs Web API MCP Search and extract web data https://t.co/goHhUmHiFo 4. web-retrieval-mcp Retrieve information across web pages https://t.co/MKVrtK5SiC 5. LlamaIndex Connect LLMs to external data https://t.co/iz62kPJguU 6. Haystack Build search and RAG pipelines https://t.co/NimEPbo6j9 7. RAGFlow Build knowledge-based AI applications https://t.co/b7T0WTDBac 8. Docling Turn documents into AI-ready content https://t.co/0qDsebf97y 9. SearXNG Run your own metasearch engine https://t.co/VMn7JbNERv 10. Crawl4AI Extract web content for AI workflows https://t.co/XvyAoWZ64A The real power isn't just generating answers. It's finding the right information, checking the sources, and connecting the dots. Bookmark this list. 🔖2d ago5.3K5325110532.70%
10 GITHUB REPOS THAT TURN THE PUBLIC WEB INTO USEFUL DATA 🌐 The internet is full of information. The real challenge is collecting it, cleaning it, and making it useful. These 10 open-source repos can help you build workflows that turn websites into structured data and actionable insights. 1. Firecrawl Crawl websites and convert pages into clean, AI-ready content. https://t.co/DySPnlefuY 2. Oxylabs Web API MCP Connect AI agents to web search, scraping, and data extraction tools. https://t.co/MG5AMRHopc 3. ScrapeGraphAI Extract information from websites using natural-language instructions and LLM-powered workflows. https://t.co/8NlrQGdfaC 4. Crawl4AI Turn web pages into structured content for AI and RAG pipelines. https://t.co/KS1t6yH8Zx 5. Newspaper4k Extract article text, metadata, and other useful information from news URLs. https://t.co/c9yoQvKkQj 6. Beautiful Soup Parse HTML and extract exactly the elements you need with Python. https://t.co/ua3bLg9w0e 7. Extruct Extract structured metadata such as JSON-LD, Microdata, and Open Graph data. https://t.co/YRi2ypSiwy 8. Pricefy Explore product-price monitoring and tracking workflows. https://t.co/y9GUBhQQU1 9. Pandas Clean, transform, and analyze collected data with Python. https://t.co/1Y6K0E0fIV 10. DuckDB Query and analyze datasets locally using a powerful analytical SQL engine. https://t.co/esAiY4UAHN From web scraping to data analysis, these tools can help turn scattered information into something useful. Collect smarter. Analyze better. Make informed decisions. Bookmark this list for your next data project.frozen6d ago5.2K471992372.19%
10 GITHUB REPOS THAT REVEAL WHAT HAPPENS BEHIND THE CODE 👀 Ever wondered what an app is doing beneath its interface? These open-source tools help you inspect binaries, explore compiled code, trace network traffic, and understand how software works internally. 1. Ghidra Reverse-engineer compiled programs and analyze binaries to uncover their internal logic. https://t.co/qj7jHQGJZi 2. JADX Decompile Android APKs into readable, Java-like source code. https://t.co/jLrIuKPU9x 3. Frida Inspect running applications by hooking functions and observing their behavior. https://t.co/NWo7FcoeEr 4. mitmproxy Capture and inspect HTTP traffic to understand how applications communicate with servers. https://t.co/euqaK2nkrZ 5. ILSpy Decompile .NET assemblies and explore their underlying C# logic. https://t.co/tHYlYb7cXB 6. APKTool Decode Android app resources and inspect Smali code inside APKs. https://t.co/207k4J2Pwo 7. Sourcegraph Search large codebases and trace functions, symbols, and dependencies across repositories. https://t.co/bgZmVxRKey 8. Semgrep Find bugs and potentially risky code patterns using customizable analysis rules. https://t.co/MBvV8y81wG 9. Tree-sitter Parse source code into syntax trees for code analysis and developer tooling. https://t.co/u6kISlvSUx 10. Binwalk Analyze firmware images and identify or extract embedded files. https://t.co/Pr0jf0M818 From understanding unfamiliar code to investigating how applications behave, these tools help you look beyond the surface. Bookmark this list. The best way to understand software is to explore what happens under the hood. Use these tools only on software and systems you're authorized to analyze.frozen6d ago5.1K421490612.48%
this guy literally built a security guard for AI coding agents. Claude Code writes code. Cursor writes code. Codex writes code. but who’s watching the agent? VibeDefend sits directly inside the coding workflow and can stop dangerous actions before they happen. it can enforce your repo rules, scan code as it’s written, protect secrets and flag security issues. one command to install: npx -y @cybedefend/vibedefend@latest install AI agents are moving faster than humans can review. this is probably a problem we’re going to see a lot more of. they also launched on Product Huntfrozen24d ago4.7K6923240172.86%
The interesting part of Base Code isn’t just AI writing code. It’s who gets to start the work. Imagine support sees the same feature request 30 times, runs it through Base Code, and hands engineering a PR instead of another ticket. The developer still decides what ships. But the backlog finally has a path to production. Try Base Code: https://t.co/phWZo5IdOAfrozen24d ago4.5K76219082.51%
An AI agent can return a 200 and still be completely wrong. The real challenge is catching failures before hundreds of users experience them. DigitalOcean’s Signals and Simulations previews help teams turn production failures into tests, validate fixes, and review them before shipping. That’s the kind of tooling AI agents need to work reliably in production.frozen10d ago4.4K9415120112.98%
10 GITHUB REPOS THAT FEEL LIKE SOMEONE LEAKED THE FUTURE A Photoshop alternative built in Rust. AI agents that help mod games. Huge AI models running with less memory. Here are 10 projects that deserve a closer look 👇 1. Photocraft A Rust-based image editor with layers, masks, and PSD support. https://t.co/FxUbiC4HAv 2. Magpie Switch between AI models for different coding agents right from your menu bar. https://t.co/4uMc4Ecrm1 3. Universal-Modder Use Claude Code to help modify PC games you own. https://t.co/Ed16PJDlTq 4. Coucou A tiny Mac menu-bar companion for keeping track of coding agents. https://t.co/C1ZgySqBXB 5. Edge0 Explore memory-efficient AI model inference by streaming model weights from your SSD. https://t.co/6mi6B2v40B 6. Shaders Add WebGPU-powered visual effects to projects built with popular frontend frameworks. https://t.co/q7O1Ya7w0z 7. Human Atlas Explore human anatomy through an interactive 3D model. https://t.co/PqhghB831b 8. Replica Skill Use Claude skills to recreate apps and refine them through user feedback. https://t.co/1NqwPLFgmn 9. Anatomy Turn descriptions of machines into interactive isometric drawings using a Claude Code skill. https://t.co/xhQYZSq130 10. Mailcheap Run your own newsletter platform with a focus on affordable email delivery. https://t.co/mhZ3qzgTC7 The best part about open source? Someone builds a weird, ambitious idea, puts it on GitHub, and suddenly anyone can experiment with it. Bookmark this list. Which project surprised you the most?2d ago4.2K261470201.59%
AI agents are moving from answering questions to actually taking actions. That makes identity, permissions, and verification much more important. Beltic is building an interesting layer for making agent capabilities more transparent and trustworthy. Excited to see how this develops.frozen22d ago4.2K6521121112.63%
10 GITHUB REPOS THAT CAN TAKE CLAUDE CODE TO THE NEXT LEVEL Claude Code is already powerful. But the right skills can make it much better at coding, debugging, testing, security, automation and real-world workflows. 1. Superpowers Adds a structured workflow for brainstorming, planning, testing, coding and review. https://t.co/2SZwrKwvCs 2. Anthropic Skills The official skills collection from Anthropic for coding, documents, testing and more. https://t.co/hz5kldPoUm 3. Matt Pocock Skills Practical skills focused on better coding habits, testing and avoiding common mistakes. https://t.co/s6gKuhT3Dj 4. Agents A large collection of specialized agents and skills for debugging, security, documentation and code review. https://t.co/XL7pBqcqHy 5. Awesome Agent Skills A huge curated collection covering development, security, testing and more. https://t.co/fMtnxU5MRN 6. Vercel Agent Skills Skills focused on React and Next.js performance, patterns and best practices. https://t.co/9x3SkMYmc7 7. Trail of Bits Skills Security-focused skills for auditing, static analysis and vulnerability detection. https://t.co/rAVt6ejjNU 8. Context Mode Helps Claude manage context during long coding sessions and stay focused on relevant information. https://t.co/HJQQE81Zq5 9. Awesome Claude Skills Connect Claude Code with tools like GitHub, Slack, Linear, Jira and many more. https://t.co/6BHMTf6peN 10. Skill Seekers Turns documentation, GitHub repos and PDFs into Claude skills automatically. https://t.co/4FCX7LalZP The interesting part isn't just Claude Code itself. It's what you can teach it to do. Claude + Skills + Context = a much more capable coding agent. Save these repos.frozen21d ago3.9K4917100533.27%
I just realized half the “apps” I want don’t need to be apps. I wanted a simple way for my friends to vote on movie night, so I asked ChatGPT to build one with Charming. A few minutes later, we had a tiny app we could actually share and use. No waiting for someone to build it. No giant project. Just make the thing you need and send it to the people who need it. Honestly, this is a pretty fun direction for AI. What tiny app would you build for your group chat? 👀 https://t.co/QhIXqV60jufrozen15d ago3.8K58169032.25%
Best 15 accounts to follow in AI: @karpathy → LLMs king @steipete → Built OpenClaw @gregisenberg → Startup ideas king @rileybrown → Vibe coding king @Shruti_0810 → Cool everday Github repos @jackfriks → Solo apps king @levelsio → Startups king @marclou → Startups king @EXM7777 → AI ops + systems king @eptwts → AI money Twitter king @godofprompt → Prompt king @vasuman → AI agents king @AmirMushich → AI ads king @0xROAS → AI UGC king @egeberkina → AI images king Follow them all and learn.frozen1mo ago3.5K9100261.04%
Real-time voice agents are getting seriously interesting. Ultravox joining Inworld could bring together strong voice technology and agent capabilities, opening up more possibilities for developers building natural voice experiences.frozen22d ago3.4K13100130.79%
This guy literally built a database that can find your files based on what’s inside them. You know that one photo you KNOW is somewhere on your phone? You search for 10 minutes. Nothing. Now imagine doing that with millions of photos, videos and documents. FEVER Multimodal DB lets you search them using normal language. “Find the video from our last event.” “Show me photos with this person.” No filenames. No folder hunting. No endless tagging. Honestly, this is how file search should work. he also launched on product hunt 👇frozen24d ago3.4K491311052.32%
Learn Computer Science for FREE: 1). HTML & CSS ↳ https://t.co/74u1K3sS8M 2). JavaScript ↳ https://t.co/UzdoEIxg32 3). DS, Algo ↳ https://t.co/qOjmKRufCg 4). Git & GitHub ↳ https://t.co/x1oQ9hhwJ2 5). React ↳ https://t.co/zc7xpFCHSh 6). SQL ↳ https://t.co/ixyjbeH596 https://t.co/Kos3xuTgqxfrozen28d ago3.1K7020160.81%
DAMN. THESE 10 GITHUB REPOS COULD HELP YOU BREAK LANGUAGE BARRIERS WITH AI. Imagine translating conversations, transcribing speech, and generating natural voices even without an internet connection. These open-source repos can help you build it: 1. Whisper Transcribe speech across multiple languages. https://t.co/BtoYkkyJno 2. Seamless Communication Build multilingual speech translation tools. https://t.co/lVRnM1Umkb 3. Argos Translate Translate text offline without relying on cloud APIs. https://t.co/iUHBteaRrX 4. Bergamot Translator Bring privacy-friendly translation directly into the browser. https://t.co/MrMA24cZXX 5. Piper Generate natural-sounding speech locally. https://t.co/wFvcXDaVE9 6. Piper Generate natural-sounding speech locally. https://t.co/wFvcXDaVE9 7. Vosk Add offline speech recognition to your apps. https://t.co/yqPxxNHibw 8. Coqui TTS Experiment with neural text-to-speech models. https://t.co/SCntx1DcHL 9. Whisper.cpp Run speech recognition efficiently on your own hardware. https://t.co/8y8dlnod7t 10. F5-TTS Generate expressive AI voices from text. https://t.co/4QpL1gNZOi The best part? These projects let developers explore speech and translation technology without building everything from scratch. Imagine creating apps that help people communicate across languages, wherever they are. Bookmark these repos. The future of communication is multilingual. 🔖1d ago2.8K431620473.84%
Developers, you’ve probably seen this happen. 👇 You ask your AI coding agent to build a feature. It writes the code. The tests pass. It says “Done.” But the feature still doesn't work the way you intended. 💀 That’s the gap @tiden_ai is solving. It verifies AI-generated code against your actual requirements, not just tests written in the same session. Works alongside Claude Code, Cursor, Codex & Copilot. Because passing tests is one thing. Building the right thing is another.frozen8d ago2.8K5423082.39%
Your phone doesn’t need another screen-time warning. It needs a moment of friction. I came across Pause recently, and the idea immediately caught my attention. Instead of simply telling you to use your phone less, Pause makes you solve a quick puzzle before opening apps like Instagram or TikTok. That tiny interruption gives you a moment to ask: “Do I actually want to scroll right now?” The founder built Pause after realizing that deleting distracting apps wasn’t solving the real problem. Delete Instagram. Reinstall Instagram. Repeat. Pause takes a different approach. You can still use social media. You just get a moment to make a conscious choice before opening it. The founder says it helped bring his Instagram usage from around 9 hours/day to less than 1 hour. Simple idea. Interesting execution. 🚀 Pause is launching on Product Hunt. If you’ve ever opened Instagram “for a minute” and somehow lost an hour, this is worth checking out. Pause. Before you scroll.frozen28d ago2.4K471411043.19%
The smartphone era might be closer to ending than we think. Meta’s AI glasses could turn your voice into the new interface. Live translation, AI assistance, and hands-free computing without pulling out your phone. The next big tech shift might not be another screen. It might be what you wear. 👀frozen5d ago2.2K7144013.62%
We've gotten pretty good at generating AI videos. The next interesting step is generating worlds you can actually interact with. Agora-2 takes that idea further by putting up to 20 humans and AI agents into the same generated environment. You don't just watch what happens. Your actions become part of what happens next.frozen28d ago2.2K3495032.32%
Small brands usually can't afford to test ads properly. Agencies and media buyers cost too much for that stage. Fastlane v2 makes, launches and iterates ads from just your website URL, and there's a 7-day free trial to test the whole loop. Access like this matters a lot for early-stage founders.11h ago2.1K1691011.31%
AI agents getting their own identity while still being tied to a real human feels like a pretty important missing piece. Instead of an app seeing 10 agents and assuming they’re 10 unrelated users, AgentID can help it understand the person actually behind them. This could make agent-based apps way easier to build and trust. kinda feels inevitable.frozen16d ago2.0K1402000.79%
Filters out generic AI-generated UI, text, and code. https://t.co/6JV7uLfYJZ https://t.co/DzgxJIUNnVfrozen14d ago1.8K1791081.93%
What if your glasses could do half the things you reach for your phone to do? Meta is getting closer to that future. A full day without a phone, testing Meta’s latest AI glasses, live translation, display features, and AI agents. Plus, Alex Himel runs his personal AI agent across 3 phones and 23 pairs of smart glasses. 👀 This is where wearable AI gets really interesting.frozen5d ago1.7K201000.17%
Planning a trip is one thing. Making it feel effortless is another. ChatGPT took 95.9 seconds to plan a day in Tokyo, while OpenUI did it in just 16.8 seconds. Both came back with photos, a map, and an itinerary, but the difference in speed changes the whole experience. Imagine having this kind of AI built right into Airbnb or https://t.co/ts9no7Y1Cp. Trip planning could get a whole lot easier.frozen7d ago1.7K551517075.56%
Every app needs a crash test dummy. Software deserves one too. Your tests can pass while real users still run into bugs nobody anticipated. That’s what makes Breken interesting. It puts your actual software into simulated production environments, throws real-world chaos at it, and catches problems before your users do. Production engineering might be the missing layer between shipping fast and shipping confidently.1d ago1.2K410000.41%