You’ve heard a lot about Umar Khalid through the media. But amid the media witch-hunt against him, there has been very little space for him to speak for himself, especially since he’s spent the last six years in jail.
So I want to tell you a little about my friend Umar, in his own words, as shared with his family and friends.
AWS Lambda built an eBPF and Rust system to log network flows across thousands of microVMs per host with zero packet drop.
Thanks to AWS Marketplace https://t.co/M0Fu98w93h
Anthropic ex-engineer runs an internal graph that costs $6 a month and catches what a $300,000 eval suite misses. No retrieval layer. Seven nodes and one rule about who's allowed to change their mind.
He published the whole schema.
His version starts from the opposite idea. A graph is not an execution order. It's a memory of why.
Seven nodes. Every edge carries the reason it exists:
> INTENT - states what the task is for. Never how
> DECOMPOSE - splits it into steps, each with a stated assumption
> WORKER - executes one step. Sees nothing else
> AUDIT - checks the output against the assumption, not the goal
> DRIFT - compares the current step to INTENT and flags divergence
> LEDGER - stores every decision with the assumption that justified it
> ROOT - holds the graph, and when an assumption breaks, re-runs every step built on it
Six nodes act. One node remembers why they acted. Every step carries the assumption that made it correct, so a false one only reruns what stood on it.
That's the entire design. A pipeline that forgets its reasons has to redo all of it or trust all of it.
He replayed a month of agent runs. 4,100 steps, 380 built on an assumption that was wrong by day three. The old pipeline shipped all 380 and linked none of them.
Everyone else builds graphs where output moves forward and the reasoning evaporates. He built one where the reason travels with the result. The blast radius is the output nobody else produces.
The article below is the full build - node prompts, the assumption format, the invalidation rule that finds every step downstream of a broken one.
Save it. You'll want it open in the other
Two Anthropic engineers, who built Claude just explained why you use less than 10% of actual Claude abilities.
This 24-minute talk will change how you use Claude Code forever.
Watch it, then read the breakdown below👇
Announcing: Agentic Document Extraction!
PDF files represent information visually - via layout, charts, graphs, etc. - and are more than just text. Unlike traditional OCR and most PDF-to-text approaches, which focus on extracting the text, an agentic approach lets us break a document down into components and reason about them, resulting in more accurate extraction of the underlying meaning for RAG and other applications. Watch the video for details.
Day 1⃣4⃣/100 What is Prompt Chaining?
Prompt chaining is a technique where multiple prompts are used in sequence, with each step refining or expanding the previous output. This improves accuracy, coherence, and reasoning in AI-generated content.
How does Prompt Chaining work?
1️⃣ Break Down the Task → Instead of solving a problem in one step, split it into multiple smaller steps.
2️⃣ Use Outputs as Inputs → The result of one prompt feeds into the next, ensuring logical flow and refinement.
3️⃣ Refine & Optimize → Intermediate outputs help guide the AI toward a well-structured and precise final result.
Let's look at an example to understand Prompt Chaining better. Imagine you want to summarize a research paper.
How would you go about it? Instead of using a single prompt to summarize, you could break the problem down into small parts via prompts and chain these prompts. Following is one way to do it:
🔹 Step 1: Extract key sections (Introduction, Methods, Results, Conclusion).
🔹 Step 2: Summarize each section separately.
🔹 Step 3: Combine the summaries into a final, polished version.
Why Use Prompt Chaining?
✅ Breaking down tasks reduces errors.
✅ Focussed prompts help keep responses structured and logical.
✅ Ideal for long-form content, reasoning tasks, and multi-step workflows.
Do you want to try prompt chaining on your favority LLM?
#AI #PromptEngineering #ChatGPT #MachineLearning #LLMs #Automation
Day 1⃣4⃣/100 What is Prompt Chaining?
Prompt chaining is a technique where multiple prompts are used in sequence, with each step refining or expanding the previous output. This improves accuracy, coherence, and reasoning in AI-generated content.
How does Prompt Chaining work?
1️⃣ Break Down the Task → Instead of solving a problem in one step, split it into multiple smaller steps.
2️⃣ Use Outputs as Inputs → The result of one prompt feeds into the next, ensuring logical flow and refinement.
3️⃣ Refine & Optimize → Intermediate outputs help guide the AI toward a well-structured and precise final result.
Let's look at an example to understand Prompt Chaining better. Imagine you want to summarize a research paper.
How would you go about it? Instead of using a single prompt to summarize, you could break the problem down into small parts via prompts and chain these prompts. Following is one way to do it:
🔹 Step 1: Extract key sections (Introduction, Methods, Results, Conclusion).
🔹 Step 2: Summarize each section separately.
🔹 Step 3: Combine the summaries into a final, polished version.
Why Use Prompt Chaining?
✅ Breaking down tasks reduces errors.
✅ Focussed prompts help keep responses structured and logical.
✅ Ideal for long-form content, reasoning tasks, and multi-step workflows.
Do you want to try prompt chaining on your favority LLM?
#AI #PromptEngineering #ChatGPT #MachineLearning #LLMs #Automation
The last 24 hours has had SO much tech news.
— Microsoft Majorana 1 quantum chip
— Google AI co-scientist
— xAI Grok 3 model release
— Clone musculoskeleton
— Claude reasoning coming soon
— Arc institute biological model
— Sakana AI CUDA kernel optimizer
All the threads here:
Watchlist, tracking your portfolio, analyzing imaginary portfolios, comparing them, getting news about stocks you own: all are in our plans. We are going to do it for sure. Just needs some time. A dirt cheap widely accessible Bloomberg Terminal for all is the dream.
Anduril is taking over IVAS, and we don't have time for business as usual.
Whatever you are imagining, however crazy you imagine I am, multiply it by ten and then do it again. I am back, and I am only getting started.
Day 1⃣3⃣/100: Let's look at reasoning models in LLMs.
1. #DeepSeekR1: Its reasoning mode is amazing, handling logical and math problems well. DeepSeek excels at showing deliberations and then choosing/calculating the correct answer.
2. #OpenAI O3-Mini: Reasoning can be accessed via API or web interface. Click "Reason" to see the thought process. O3-mini does not show deliberations but shows the reasoning for the final answer.
3. Google Gemini Flash Thinking: An experimental product, less effective. Available in Google AI Studio, it makes math problems more academic and convoluted.
What do you think? Reply with your opinions.
Follow for more and check out "Prompt Engineering in Action" for prompt engineering insights.
@ManningBooks https://t.co/bX812HtTZB
#PromptEngineering #AI #PromptEngineeringInAction