X / Twitter
Our kids are 8 and 10; Here are some of the parenting tools and resources we loved this year:
https://t.co/vgoOHYglt3 https://t.co/wLWKxthTJY
a comprehensive guide on AI Agents
just published by @i_amanchadha:
> agentic design patterns
> function, tool, API calling
> planning and multi-agent collaboration
> agentic RAG and code implementation
> benchmarks and frameworks
> related papers https://t.co/J35dMihptg
MAKING SOFTWARE
A reference manual for people who design and build software.
→ https://t.co/IufmYESQwk https://t.co/KHfTi6g6Ph
With the exciting new MCP integration with @windsurf_ai, we’ve creating a tutorial to guide you through connecting Windsurf to @supabase in 4 easy steps!
See it in action below 👇 https://t.co/uPBsCjmJtC
Use ForeverVM with Claude Desktop to perform complex data analysis or draw a mandlebrot with ASCII characters.
Introducing the ForeverVM MCP server for code execution in a Python REPL: https://t.co/1yzkHNJL2v https://t.co/8ge0q01bum
Ultimate MCP tutorial 🤯🤯🤯
Learn how to configure MCP in Cursor, Windsurf and Claude
In this tutorial, we used the github mcp server
a thread 🧵👇 https://t.co/fDE49ExE0c
We're excited to launch Windsurf Directory together with @windsurf_ai, a place where you can:
◇ Find curated rules
◇ Explore MCPs
◇ Learn from videos
◇ Join the community
Link ⬇️🧵 https://t.co/W34JfXnO2R
if you want something that does work, spend money on private tutoring instead of private schools https://t.co/jsmxS3zoG3 https://t.co/9yiRML5z6j
Hot take: The money you save by NOT sending your kids to private school will benefit them more as a lump sum in the future than the private school education would have.
I highly recommend brand owners on here start using Gumloop for ad production. It’s like Zapier for AI.
This flow scrapes any brand’s ad library, transcribes its video ads, repurposes the scripts for your brand, and puts them in a Google sheet.
Took me 30 seconds to build. https://t.co/zpFFoAbomI
If you're still having questions about MCP and its potential use cases, here's a quick demo on how MCP can work within Cascade! https://t.co/TbgJExdVdc
I wrote this up mainly to illustrate my workflow for LLM-based exploratory prototyping - being able to get from a loose idea to a working prototype in 15 minutes is such a productivity boost https://t.co/3FdcPYEDY2
Introducing apple-mcp - https://t.co/HEAPFOu8ZK
one simple command to give LLMs access to a bunch of apple-native tools like
- contacts
- notes
- iMessages
and more (soon)
just add this to your claude desktop config. done. https://t.co/0otiUYcXcB
Fiverr just launched a platform for freelancers to train a custom AI model, and "sell" the ability to generate work with it.
They're pitching it as a way for creators to scale the number of projects they can handle at once.
Models are trainable on text, image, and sound. https://t.co/CP5Kdr7Sdv
AI Agents 101: 𝗔𝗜 𝗔𝗴𝗲𝗻𝘁 𝗠𝗲𝗺𝗼𝗿𝘆.
In general, the memory for an agent is something that we provide via context in the prompt passed to LLM that helps the agent to better plan and react given past interactions or data not immediately available.
It is useful to group the memory into four types:
𝟭. Episodic - This type of memory contains past interactions and actions performed by the agent. After an action is taken, the application controlling the agent would store the action in some kind of persistent storage so that it can be retrieved later if needed. A good example would be using a vector Database to store semantic meaning of the interactions.
𝟮. Semantic - Any external information that is available to the agent and any knowledge the agent should have about itself. You can think of this as a context similar to one used in RAG applications. It can be internal knowledge only available to the agent or a grounding context to isolate part of the internet scale data for more accurate answers.
𝟯. Procedural - This is systemic information like the structure of the System Prompt, available tools, guardrails etc. It will usually be stored in Git, Prompt and Tool Registries.
𝟰. Occasionally, the agent application would pull information from long-term memory and store it locally if it is needed for the task at hand.
𝟱. All of the information pulled together from the long-term or stored in local memory is called short-term or working memory. Compiling all of it into a prompt will produce the prompt to be passed to the LLM and it will provide further actions to be taken by the system.
We usually label 1. - 3. as Long-Term memory and 5. as Short-Term memory.
A visual explanation of potential implementation details 👇
And that is it! The rest is all about how you architect the topology of your Agentic Systems.
What do you think about memory in AI Agents?
#LLM #AI #MachineLearning
I think this is pretty interesting:
Fine-tuning a model for function calling.
When it comes to building agent systems, we often talk about which massive model is the most powerful at reasoning.
But sometimes, you might not need a model that excels at reasoning.
Sometimes, you might just want a model that excels at calling functions.
In this bonus unit of Hugging Face's Agents course, you'll learn:
- how to fine-tune Gemma 2 2b IT
- using the hermes-function-calling-v1 dataset by NousResearch
- using LoRA fine-tuning
- and all within 6 hours in Google Colab with the free tier T4 GPU.
Pretty cool.
Here's the link to the notebook:
https://t.co/YoaxTfa8Rk
FLARE: Feed-forward Geometry, Appearance and Camera Estimation from Uncalibrated Sparse Views
Contributions:
- We propose an efficient, feed-forward, and differentiable system for high-quality 3D Gaussian scene reconstruction from uncalibrated sparse-view images, achieving inference in less than 0.5 seconds.
- We demonstrate that leveraging camera poses as proxies effectively simplifies complex 3D learning tasks. We introduce a novel cascaded learning paradigm that starts with camera pose estimation, whose results condition the subsequent learning of geometric structure and appearance.
- We propose a two-stage geometry learning approach that first learns camera-centric pointmaps and builds a global geometry projector to unify the pointmaps into a global coordinate.
A beginner's guide to how to use MCP! https://t.co/HLC0Vh723Y
Transform ANY Content into Clean Markdown with Markdownify MCP Server
Think about the last time you struggled with file conversions. Now imagine:
File Conversions:
- PDFs → Markdown
- Images → Text
- Audio → Transcripts
- Office docs → Clean text
Web Content:
- YouTube videos
- Search results
- Web pages
- Existing Markdown
Why it matters:
• Single solution for multiple formats
• Clean, readable output
• Simple integration
• Developer-friendly setup
The beauty isn't just WHAT it converts, but HOW EASILY it does it.
Ready to transform your content? Three steps using Cline:
1. Clone the repo
2. Install dependencies
3. Build & Run
Remember: Great tools don't just solve problems—they eliminate them entirely.
Here’s how to build your app idea using OpenAI o1-pro with my 6-prompt template system.
Even if you can’t code.
Watch for a 3hr 58min tutorial and learn how to build + code with AI.
Prompts & links below. https://t.co/U0bAe0TUx1