THIMIRA NIRMAL

Love science technology and history

[THIMIRA NIRMAL @TIMNIRMAL]

If anyone needs starlink referel code for free month 🚀😉

https://t.co/wxuhdUcgCn

[THIMIRA NIRMAL @TIMNIRMAL]

Things are coming live 🤩

Human chat is done, now to connecting Agents

[THIMIRA NIRMAL @TIMNIRMAL]

I don't think agents mainly have a planning problem anymore.

They have an execution problem.
The models are getting very good at deciding what should happen.

Keeping reality consistent with that plan is a completely different engineering problem.

#ai #agents #aiagents #chatgpt #claude

[THIMIRA NIRMAL @TIMNIRMAL]

Still building the UI for AgentRuntime Space 👀

The idea is bigger than putting an AI assistant inside a workspace.

Agents in Space can talk to people.
They can talk to other agents.
And importantly, they don’t always have to wait for someone to prompt them — they can initiate conversations and actions too.

Behind that, they can:

> Run AgentRuntime workflows
> Use our MCP/connectors
> Work with files and docs
> Collaborate with other agents and people
> Operate under the same workspace permissions and usage based billing without drama

Basically, bring your people, agents, workflows, tools, docs and files into one place — and let them actually engage with each other.

Kinda like the idea of a bot living inside your network…

except the bot can actually do the work. 😉

App goes into internal testing this week.

#AgentRuntime #AIAgents #AgenticAI #FutureOfWork

[THIMIRA NIRMAL @TIMNIRMAL]

𝐀𝐠𝐞𝐧𝐭𝐫𝐮𝐧𝐭𝐢𝐦𝐞 𝐒𝐩𝐚𝐜𝐞 𝐚𝐧𝐝 𝐖𝐨𝐫𝐤𝐛𝐞𝐧𝐜𝐡.

Runtime and Console are already out there.
Now two more things are starting to take shape

No explanation for now 👀

Just looking at the names and these screenshots:

What do you think Space is?
What do you think Workbench is?

Curious what these names make you expect before I tell you what we’re actually building.

#AgentRuntime #AIAgents #AgenticAI #DeveloperTools @agentruntime

[THIMIRA NIRMAL @TIMNIRMAL]

This is starting to feel pretty real.

AgentRuntime MCP is coming soon.

You’ll be able to create workflows, run them, update runtime context, handle approvals, inspect executions and more — directly from ChatGPT.

No jumping between tools just to operate the workflow.

In this test, ChatGPT changed the context for the current run and approved an existing human approval gate, while AgentRuntime still controls the actual execution underneath.

The chat becomes the interface.
AgentRuntime becomes the runtime.
More soon.

#AgentRuntime #MCP #AIagents #AgenticAI #Automation

[THIMIRA NIRMAL @TIMNIRMAL]

1.5 B token after...

[THIMIRA NIRMAL @TIMNIRMAL]

🥵☕

[Mitchell Hashimoto @mitchellh]

Cool to see Gemini CLI, but there isn't a more dysfunctional product, pricing, and packaging organization in the entire world than Google, so that's my primary hesitation... I had to get a Gemini API key recently and it was a reminder of how bad at this Google is.

[THIMIRA NIRMAL @TIMNIRMAL]

🌐 Inputs: files, folders, URLs.
📊 Extracts structured data easily.
🛠️ Usage:
processor.process_file(file_path)
processor.process_folder(TEST_FOLDER, OUTPUT_FOLDER)

✨ Early release – Contributions welcome!
#AI #MachineLearning #DocumentProcessing

[THIMIRA NIRMAL @TIMNIRMAL]

🚀 gemini-parser – AI-powered document parsing with Google Gemini API!

📜 Parses large PDFs with chunking & parallel processing.
⚡ Smart caching for speed & lower costs.
📑 Supports PDF, CSV, HTML, DOC, XML, TXT.
🔗 GitHub: https://t.co/LuuHY2cZKm

#AI #GoogleGemini #Python

[OpenAI @OpenAI]

Deep research is now rolled out to Pro users.

We are still working on bringing access to users in the UK, EU, Norway, Iceland, Liechtenstein, and Switzerland.

[THIMIRA NIRMAL @TIMNIRMAL]

✅ Features
Drag-and-drop nodes 🖱️
Interactive connections 🔗
Dark mode toggle 🌙
Custom context menu 🎛️

No paid libraries, no extra fluff—just clean, efficient, and elegant code. ✨ 💡 Want to learn how I built it?

Check out the detailed guide here: https://t.co/VWFmz93V6s

[THIMIRA NIRMAL @TIMNIRMAL]

🚀 Custom Workflow Builder / Node base editor! 🌟

Tired of complex and expensive libraries for workflow creation? I built my own lightweight, customizable Workflow Builder using React and TailwindCSS! 🎨

💡 Want to learn how I built it?
Github Code:
https://t.co/ar0U765lUa

[THIMIRA NIRMAL @TIMNIRMAL]

@OpenAI Realtime API in Work 🤪 + @langchain Tools + RAG

[THIMIRA NIRMAL @TIMNIRMAL]

🚀 YOLO 11 Has Arrived! Redefining What's Possible in AI! 🚀

🌟 @ultralytics YOLO 11 is here, and it’s faster, smarter, and more powerful than ever! Whether you’re working on autonomous driving, healthcare, or smart retail, YOLO11 is designed to take your computer vision projects to the next level! 🔥

✨ Key Features of YOLO 11:

🔍 Object Detection: Pinpoints objects in images & videos for tasks like surveillance, autonomous driving, and retail.
🖼️ Instance Segmentation: Breaks down images to pixel-level detail for medical imaging and manufacturing.
🏷️ Image Classification: Automatically tags images for e-commerce, wildlife monitoring, and more.
🏋️ Pose Estimation: Tracks movements for fitness, sports analytics, and healthcare applications.
🛰️ Oriented Object Detection (OBB): Spots objects at an angle — perfect for aerial imagery, robotics, and warehouse automation.
🎥 Object Tracking: Keeps track of objects across frames for real-time security and traffic management.

🔧 Why YOLO11 is a Game-Changer:

⚡ Faster Processing: 2% quicker than YOLOv10 — perfect for real-time applications!
🎯 Higher Accuracy: Better feature extraction with 22% fewer parameters, delivering more accurate results.
🌍 Efficient & Flexible: Works on both cloud and edge devices, making it ideal for all environments.
💡 Easy Integration: Plug it into your existing systems with support for platforms like NVIDIA GPUs and Google Colab.

💥 Why YOLO11?

⚙️ Speed, accuracy, and efficiency combined like never before.
🎯 Perfect for industries like healthcare, agriculture, smart retail, and autonomous vehicles.
🌐 Backed by Ultralytics HUB, offering cloud training, pre-trained models, and community support!
Join the AI revolution with YOLO11 and unlock new possibilities for your vision projects! 🚀💡

👉 Get Started Today: https://t.co/LuGwmjioD9

Models Will be available from September 30

#YOLO11 #AI #ComputerVision #ObjectDetection #MachineLearning #Ultralytics #TechInnovation #SmartAI #DeepLearning #VisionAI

[THIMIRA NIRMAL @TIMNIRMAL]

Meet Llama 3.2: AI Evolved! 🌟

Now with Vision Capability - See and understand the world like never before! Plus, introducing Llama Guard for safer, smarter interactions. Your digital world just got a whole lot brighter! 🐪✨

#Llama3.2 #AIVision #LlamaGuard #NextGenAI #TechUpdates

[Runway @runwayml]

Gen-3 Alpha Video to Video is now available on web for all paid plans. Video to Video represents a new control mechanism for precise movement, expressiveness and intent within generations. To use Video to Video, simply upload your input video, prompt in any aesthetic direction you like, or, choose from a collection of preset styles.

See examples below.

(1/6)

[THIMIRA NIRMAL @TIMNIRMAL]

🚀 EVI 2 – The Voice AI Everyone’s Talking About! 🎤💬

Just came across this incredible new voice-to-voice AI model from Hume AI called EVI 2—and I had to share it with you all! 🤖✨ If you’re into AI, this one’s worth checking out.

🔥 What Makes EVI 2 So Cool?

⚡️ Fast and Fluent: Real-time responses that feel super natural.
🌍 Multilingual Skills: Easily switch between different languages.
🎭 Customizable Voices: You can adjust everything from pitch to accents!
🎤 Emotionally Intelligent: It even picks up on your tone of voice for more engaging conversations.

Oh, and the best part? No voice cloning risks—it’s all about security and unique experiences.

Want to try out?
https://t.co/zVL9qP4XfX

Definitely worth a read if you're curious about where voice AI is heading!

#AI #VoiceAI #TechInnovation #HumeAI #EVI2 #AIConversations #TechNews

[THIMIRA NIRMAL @TIMNIRMAL]

🍓 OpenAI strawberry is out 🍓

Introducing OpenAI o1 – a breakthrough in AI reasoning! 🤖💡

OpenAI has just unveiled its new series of models, designed to solve complex problems by thinking deeper before responding. Whether you're tackling difficult coding challenges, scientific research, or mathematical puzzles, OpenAI o1 is here to help!

🔬 What makes it special?

> Trained to reason through problems like a person, improving accuracy and problem-solving.
> Outperforms previous models, solving 83% of difficult problems in the International Mathematics Olympiad and achieving impressive coding scores in Codeforces competitions.
> Safer and more aligned, thanks to enhanced safety measures and reasoning-based safeguards.

💻 Who can benefit?

Developers, scientists, researchers, and anyone facing complex challenges in areas like healthcare, physics, coding, and more!

💡 Bonus: OpenAI o1-mini is also here! 80% cheaper, faster, and perfect for coding tasks without sacrificing reasoning power.

Try it out now on ChatGPT or via the API! 🌐

#OpenAIo1 #OpenAI