https://xitter.cf/t.co/yZfMHuNNta
Maringá/PR - Brazil
Joined September 2009
- Tweets3.7K
- Following1.9K
- Followers375
- Likes3.5K
Clayton Passos retweeted
AI Reverse Engineer 'GOD OF WAR' HD now runs fully natively on any Windows Pc.
- Original PS4 code running directly on your CPU.
- Average 144 FPS at 1080p.
- playable start to finish, up to 240 FPS engine rate.
- Not a full PS4 emulator a game-specific runtime.
- github.com/FelipePrado0/gow3…
GTA-,V is running in a browser on at 60 FPS.
it back on browser, No download, No launcher.
- Ai is cooking really hard
- gta.deloskiyt.pl
Clayton Passos retweeted
Jev is TypeSafe AI's first System One Model.
It is 100x faster and cheaper than frontier LLMs.
That opens up a lot of use cases people usually skip because the big models are too slow or too expensive for them.
Here are 9 use cases where we think Jev could be used instead of an LLM.
Clayton Passos retweeted
3 agent patterns are dominating AI systems right now:
1. Planner → Executor
One agent creates the plan. Another carries it out.
2. Router → Specialist
One agent decides whether the task needs a debugger, security reviewer, test writer, or another expert.
3. Map → Reduce
Several agents work on smaller parts in parallel. Another combines the results.
We have seen these before in distributed systems, and they share the same challenge:
Each agent needs enough context to do its job, but not so much it gets confused.
That means defining:
- Where one agent’s responsibility ends
- What the next agent needs
- How results should be validated
- Who owns the final decision
And still, most "agent failures" I'm seeing so far aren't model failures; they're simply context failures.
Thoughts?
Clayton Passos retweeted
5 LLM quantization techniques, clearly explained:
(bookmark this)
A 70B model in FP16 needs 140GB for weights alone. At 4-bit, that drops to 35GB, which fits on one card.
But naive rounding fails on large models. Roughly 0.1% of hidden dimensions carry values up to 20x larger than anything else in the tensor, and they wreck the quantization grid for everything else.
Each of these 5 methods handles those outliers at a different point:
1. RTN: ignores them. Rounds every weight to the nearest grid level with no calibration data. Cheapest option, weakest at low bit widths.
2. GPTQ: repairs after rounding. Quantizes a layer column by column and adjusts the remaining weights to absorb the error before moving on.
3. AWQ: protects before rounding. Finds the ~1% of weight channels that matter most and scales them up so they survive quantization. Everything still ends up in plain INT4.
4. LLM. int8(): isolates at inference. Outlier dimensions run in FP16, the other 99.9% run in INT8, and the results are merged.
5. QAT: solves it during training. The model is fine-tuned with rounding baked into every forward pass, so it adapts to the damage before quantization is actually applied.
All five produce the same artifact, a model at a fraction of its trained precision. They differ only in where the outlier problem gets addressed.
The visual below nicely summarises these techniques.
As further reading, the article below is a first-principles guide to LLM inference that walks through everything between your prompt and the streamed response, covering tokenization, embeddings, attention, the prefill and decode split, KV caching, and quantization.
It will give you a complete mental model of how inference actually works under the hood.
Read it below.
Clayton Passos retweeted
Vector Database by hand ✍️ ~ 10 steps walkthrough below
Vector databases are the backbone of Retrieval Augmented Generation (RAG).
How do they actually work?
Goal: index three sentences, then answer a query by finding the nearest one, filling in every cell yourself.
= 1. Given =
A dataset of three sentences, three words each. In practice it is millions of them.
= 2. Word embeddings =
Let us look up each word in an embedding table. Here the vocabulary is 22 words; in practice it is tens of thousands, and the vectors have thousands of dimensions rather than four.
= 3. Encoding =
We feed the sequence to an encoder, one linear layer and a ReLU, and get one feature vector per word. In practice the encoder is a transformer.
= 4. Mean pooling =
Let us average across the columns. Three word vectors collapse into one, which is what people mean by a text embedding or a sentence embedding.
= 5. Indexing =
We multiply by a projection matrix and the four dimensions become two. It is doing the job of a hash: a short representation that is faster to compare, and it is what gets saved in the vector storage.
= 6. Process "who are you" =
Let us repeat steps 2 to 5 on the second sentence.
= 7. Process "who am I" =
We do it a third time. The database is now indexed.
= 8. Query "am I you" =
Let us push the query through the very same pipeline: lookup, encoder, mean pooling, projection, and it lands as a 2D vector in the same space.
= 9. Dot products =
We transpose the query and multiply, which takes the dot product against every stored vector at once. The dot product is the estimate of similarity.
= 10. Nearest neighbour =
Let us scan for the largest: 60/9 beats 44/9 and 40/9, so the answer is "who am I". Scanning billions of vectors one at a time is what makes this the slow step in practice, which is why real databases use an approximate nearest neighbour index like HNSW.
The outputs:
Stored index vectors = [5/3, 2/3], [5/3, 0], [7/3, 2/3]
Query vector = [8/3, 2/3]
Dot products = 44/9, 40/9, 60/9
Nearest neighbour = "who am I"
The takeaway: a vector database is an embedding pipeline, a projection, and a dot product. Every step here is arithmetic you can do in pen, which is worth remembering when the word "database" makes it sound like something else.
💾 Save this post!
Clayton Passos retweeted
RAG is not just one technique, it is an entire ecosystem of intelligence.
From context-aware assistants to domain-specific systems, here are 16 types of RAG models shaping the next wave of AI innovation -
1. Standard RAG
The foundation of all RAG systems - combines retrieval and generation for question answering and knowledge synthesis.
2. Agentic RAG
Empowers AI agents to retrieve and act autonomously, perfect for assistants that need dynamic, tool-based reasoning.
3. Graph RAG
Uses knowledge graphs for relational reasoning - ideal for expert systems in law, medicine, and semantic search.
4. Modular RAG
Breaks retrieval, reasoning, and generation into independent components - enabling collaborative, scalable AI workflows.
5. Memory-Augmented RAG
Adds persistent external memory for context retention, powering long-term chatbots and personalized experiences.
6. Multi-Modal RAG
Processes text, images, and audio together - perfect for video summarization, captioning, and multi-modal AI tools.
7. Federated RAG
Enables privacy-preserving retrieval from decentralized sources, used in healthcare and secure enterprise systems.
8. Streaming RAG
Performs real-time retrieval and generation, ideal for financial dashboards, live feeds, and social media monitoring.
9. ODQA RAG (Open-Domain QA)
Handles large, diverse datasets - ideal for search engines and intelligent virtual assistants.
10. Contextual Retrieval RAG
Maintains session-level awareness, great for conversational AI and customer support chatbots.
11. Knowledge-Enhanced RAG
Integrates structured domain data, useful for legal, educational, and professional knowledge applications.
12. Domain-Specific RAG
Custom-tailored for specific industries - like finance, healthcare, or legal analytics.
13. Hybrid RAG
Combines multiple retrieval approaches, bridging structured and unstructured data for high precision.
14. Self-RAG
Introduces self-reflection to refine its own answers, enabling AI models to fact-check and improve reasoning autonomously.
15. HyDE RAG (Hypothetical Document Embeddings)
Generates hypothetical documents to guide retrieval, excellent for complex or niche query contexts.
16. Recursive / Multi-Step RAG
Performs multiple retrieval-generation loops, enabling advanced problem-solving and reasoning chains.
From simple retrievals to self-improving AI reasoning loops, RAG is evolving fast.
Which type do you think will dominate enterprise AI systems in 2026?
Clayton Passos retweeted
A imprensa brasileira tem falhado ao informar, de forma acrítica e sem qualquer ponderação, o ingresso de novos países no banimento de menores de idade às redes sociais.
No verso do discurso que veículos têm reproduzido das assessorias de imprensa desses governos está o método desse banimento: a coleta massiva de dados pessoais e biométricos de adultos a cada acesso a serviços online para provarem não serem menores de idade. Estão nos mostrando só a folha da frente.
Portanto, toda notícia de restrições online a crianças e adolescentes que omita a informação de que os maiores afetados (em número e extensão de impacto) são os adultos entrega uma informação incompleta, merecendo críticas de seus leitores e espectadores.
A repetição na íntegra de um discurso governamental sem análise, sem crítica e sem ponderação chama-se publicidade. E esta deve ser cobrada pelo veículo e identificada ao leitor como tal.
Clayton Passos retweeted
6 anos usando nubank, 5 anos de ultravioleta, nunca devi 1 real pro banco e sempre tive um valor alto investido
Acordei hj com minha conta encerrada definitivamente e com todo saldo preso sem nem uma explicação
Literalmente me vi no meme do pai mandando o filho ir na agência…
Clayton Passos retweeted
⚠️ Malicious Sicoob NuGet steals Brazilian bank credentials while npm packages target AWS and CI/CD secrets.
The fake "Sicoob.Sdk" versions 2.0.0–2.0.4 exfiltrate client IDs, PFX certificates, and passwords. It was downloaded nearly 500 times.
Multiple npm packages from one actor also steal cloud and pipeline secrets.
Full report: thehackernews.com/2026/05/ma…
Clayton Passos retweeted
Save this post right now.
This is the exact 9-step system to create a professional product video with Claude.
No agency. No video team. No $5,000 production budget. Just you and Claude.
Here is exactly how:
☑️ 1. What You Need Before You Start
→ One clean product photo, white or neutral background preferred.
→ A Claude Pro account for MCP connections.
→ One clear goal picked before you touch anything: awareness, conversion, or retargeting.
↳ Clean photo in = clean video out. This step alone saves 30 minutes of fixing.
☑️ 2. Connect Higgsfield MCP to Claude
→ Go to Claude. ai > Connectors → Search "Higgsfield." → Click Add.
→ Authenticate and return to Claude, Higgsfield tools now appear in your chat.
→ Type "list available Higgsfield tools" to confirm the connection is live.
↳ This unlocks Claude, sending your product image directly to Higgsfield and retrieving the video.
☑️ 3. Write Your Video Brief With Claude
Paste this prompt:
"I am creating a 15-second TikTok product video for [product]. Target audience: [audience]. Tone: [energetic/luxurious/cinematic]. Goal: [awareness/conversion]. Write: video concept in 2 sentences, 3-scene shot sequence, voiceover script under 15 seconds, text overlay plan for each scene."
↳ The brief directs everything. Skip it, and your video is random.
☑️ 4. Generate the Product Video With Higgsfield MCP
Prompt:
"Using Higgsfield, generate a product video for the image I will upload. Motion: [slow zoom/orbit/floating/cinematic pan]. Style: [luxury/energetic/minimal]. Duration: 15 seconds. Subtle depth of field. Output in 9:16 vertical format."*
↳ Motion style guide: Slow zoom = luxury. Orbit = tech. Floating = wellness. Cinematic pan = fashion.
☑️ 5. Write the Voiceover With Claude
Prompt:
"Write a 15-second voiceover for [product]. Tone: [tone]. Hook in the first 2 seconds. Short, punchy sentences. End with soft CTA: 'Shop now' or 'Try it today.' Max 40 words total."
↳ Voice style by product: Luxury = slow, breathy. Tech = clear, fast. Wellness = warm, calm. Fashion = punchy, bold.
☑️ 6. Add Text Overlays and Captions With Claude
Prompt: "Write a text overlay plan for a 15-second video for [product]. Divide into 3 scenes, each 5 seconds long. For each scene: overlay text (max 5 words), position, font style, and timing."
→ Add auto-captions and apply one color grade preset matching the brand's mood.
☑️ 7. Write the Caption and Hashtags With Claude
Caption prompt:
"Write 3 TikTok captions for [product]. Each: hook under 10 words, soft CTA, under 150 chars. Label them: Curiosity / Social Proof / Benefit-led."
For more AI systems like this: 👇
→ Go to substack.com/@humzakhalid
→ Subscribe to my free newsletter (don't pay anything)
→ Get more free and daily cheatsheets
----
Which product are you making your first video for? Drop it below.
----
♻️ Repost to give your network an unfair advantage.
Clayton Passos retweeted
𝗠𝗖𝗣 plus 𝗔𝟮𝗔, here is how they complement each other 👇
Protocol wars continue to rage, let's understand how Googles A2A (Agent2Agent) protocol is different from MCP and how they complement each other (read till the end).
𝘔𝘰𝘷𝘪𝘯𝘨 𝘱𝘪𝘦𝘤𝘦𝘴 𝘪𝘯 𝘔𝘊𝘗:
𝟭. MCP Host - Programs using LLMs at the core that want to access data through MCP.
❗️ When combined with A2A, an Agent becomes MCP Host.
𝟮. MCP Client - Clients that maintain 1:1 connections with servers.
𝟯. MCP Server - Lightweight programs that each expose specific capabilities through the standardised Model Context Protocol.
𝟰. Local Data Sources - Your computer’s files, databases, and services that MCP servers can securely access.
𝟱. Remote Data Sources - External systems available over the internet (e.g., through APIs) that MCP servers can connect to.
𝘌𝘯𝘵𝘦𝘳 𝘈2𝘈:
Where MCP falls short, A2A tries to help. In multi-Agent applications where state is not necessarily shared
𝟲. Agents (MCP Hosts) would implement and communicate via A2A protocol, that enables:
➡️ Secure Collaboration.
➡️ Task and State Management.
➡️ User Experience Negotiation.
➡️ Capability discovery - similar to MCP tools.
𝗛𝗼𝗻𝗲𝘀𝘁 𝘁𝗵𝗼𝘂𝗴𝗵𝘁𝘀:
❗️ Open protocols for Agent communication are important but they are just a piece of the picture, there will be a need for standards that govern all of the existing protocols and other missing pieces. E.g.
❓ How do we standardise tracing and Observability in multi-agent IOA systems?
❓ How do we retain the identity of the running job of an AI Agent instance if the communication standard is not unified in different parts of the pipeline?
❓ ...
✅ More on this in my future posts, stay tuned!
Let me know your thoughts in the comments. 👇
Clayton Passos retweeted
You must know these 𝗔𝗴𝗲𝗻𝘁𝗶𝗰 𝗦𝘆𝘀𝘁𝗲𝗺 𝗪𝗼𝗿𝗸𝗳𝗹𝗼𝘄 𝗣𝗮𝘁𝘁𝗲𝗿𝗻𝘀 as an 𝗔𝗜 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿.
If you are building Agentic Systems in an Enterprise setting you will soon discover that the simplest workflow patterns work the best and bring the most business value.
At the end of last year Anthropic did a great job summarising the top patterns for these workflows and they still hold strong.
Let’s explore what they are and where each can be useful:
𝟭. 𝗣𝗿𝗼𝗺𝗽𝘁 𝗖𝗵𝗮𝗶𝗻𝗶𝗻𝗴: This pattern decomposes a complex task and tries to solve it in manageable pieces by chaining them together. Output of one LLM call becomes an output to another.
✅ In most cases such decomposition results in higher accuracy with sacrifice for latency.
ℹ️ In heavy production use cases Prompt Chaining would be combined with following patterns, a pattern replace an LLM Call node in Prompt Chaining pattern.
𝟮. 𝗥𝗼𝘂𝘁𝗶𝗻𝗴: In this pattern, the input is classified into multiple potential paths and the appropriate is taken.
✅ Useful when the workflow is complex and specific topology paths could be more efficiently solved by a specialized workflow.
ℹ️ Example: Agentic Chatbot - should I answer the question with RAG or should I perform some actions that a user has prompted for?
𝟯. 𝗣𝗮𝗿𝗮𝗹𝗹𝗲𝗹𝗶𝘇𝗮𝘁𝗶𝗼𝗻: Initial input is split into multiple queries to be passed to the LLM, then the answers are aggregated to produce the final answer.
✅ Useful when speed is important and multiple inputs can be processed in parallel without needing to wait for other outputs. Also, when additional accuracy is required.
ℹ️ Example 1: Query rewrite in Agentic RAG to produce multiple different queries for majority voting. Improves accuracy.
ℹ️ Example 2: Multiple items are extracted from an invoice, all of them can be processed further in parallel for better speed.
𝟰. 𝗢𝗿𝗰𝗵𝗲𝘀𝘁𝗿𝗮𝘁𝗼𝗿: An orchestrator LLM dynamically breaks down tasks and delegates to other LLMs or sub-workflows.
✅ Useful when the system is complex and there is no clear hardcoded topology path to achieve the final result.
ℹ️ Example: Choice of datasets to be used in Agentic RAG.
𝟱. 𝗘𝘃𝗮𝗹𝘂𝗮𝘁𝗼𝗿-𝗼𝗽𝘁𝗶𝗺𝗶𝘇𝗲𝗿: Generator LLM produces a result then Evaluator LLM evaluates it and provides feedback for further improvement if necessary.
✅ Useful for tasks that require continuous refinement.
ℹ️ Example: Deep Research Agent workflow when refinement of a report paragraph via continuous web search is required.
𝗧𝗶𝗽𝘀:
❗️ Before going for full fledged Agents you should always try to solve a problem with simpler Workflows described in the article.
What are the most complex workflows you have deployed to production? Let me know in the comments 👇
Clayton Passos retweeted
Anthropic's Code With Claude Announces Managed Agents, Proactive Workflows, Capability Curve bit.ly/4nBADE9 by @sithypoo
Clayton Passos retweeted
Most people use Claude like a chatbot.
The top 1% use it like a system.
The people actually winning with AI in 2026 have given Claude a job title, a file, and a system.
Here are the 20 Claude Skills everyone needs this year:
Here is exactly how: 👇
☑️ 1. Content and Growth
→ viral-hook. MD generates 10 scroll-stopping opening lines for any niche.
→ brand-voice. MD trains Claude on your tone so it never sounds like AI
→ repurpose. MD turns one YouTube script into 5 LinkedIn posts instantly.
↳ seo-expert. MD and audience-pains. MD, round this out. Your content engine, automated.
☑️ 2. Executive and Admin
→ meeting-pro. MD converts messy transcripts into actionable tasks.
→ calendar-shield. MD audits and finds 4 hours of hidden deep-work time.
→ critic-loop. MD acts as a harsh Advocate to find every flaw in your plan.
☑️ 3. Tech and Systems
→ code-auditor. MD scans for security holes and Clean Code violations.
→ unit-tester. MD writes a full test suite for every new function you create.
→ sql-master. MD converts plain English into complex database queries.
☑️ 4. Strategy and ROI
→ market-gap. MD analyzes 2 competitors and finds their weaknesses.
→ roi-projector. MD forecasts revenue based on conversion and ad spend.
→ vision-check. MDensures every task you do today aligns with your 12-month business goal.
I wrote the full breakdown on how to build every single one of these skills yourself. Step-by-step. Copy-paste ready.
Read it here → open.substack.com/pub/humzak…
For more Claude systems like this: 👇
→ Go to substack.com/@humzakhalid
→ Subscribe to my free newsletter (don't pay anything)
→ Get more free and daily cheatsheets
Which skill are you installing first? Drop the number below, and I'll share the exact prompt to activate it.
♻️ Repost to give your network an unfair advantage.
Clayton Passos retweeted
Caching Strategies in System Design: Types, Patterns, Trade-offs & Best Practices buff.ly/BNd5XRc
Clayton Passos retweeted
Most developers spend 2 weeks onboarding to a new AI coding tool.
This cheatsheet cuts it to 24 hours.
4 sprints. 1 loop. Zero wasted time.
Here is the full breakdown: 👇
☑ 1. Setup
→ Install Claude Code, connect the VS Code extension, and sign in.
→ Link exactly 3 repos. Not 10. Not your entire org.
→ Start with a minimal, organized codebase.
☑ 2. The TDD Core Loop
→ Write the test before you write any code.
→ Hand Claude this prompt: "Given this feature: [describe]. Write a Python test suite with 10+ Pytest cases. Then write the implementation that passes every test completely."
→ Run the test. If it fails, loop. If it passes, move to step 3.
☑ 3. Context Handling for Legacy Repos
→ Repo over 500 files? Stop. Use a specialized Context Agent.
→ Prompt: "Analyze this 10-year-old JavaScript file: [content]. Find 5 refactoring opportunities that reduce complexity without breaking existing logic."
→ Establish a Rule of 3: only the three most relevant files per session.
☑ 4. Deploy and Lock It In
→ Connect GitHub, AWS Lambda, Docker, and VS Code via connectors
→ Generate a GitHub Actions YAML file that runs Pytest on every commit.
→ Route coverage reports into PR comments.
Manage context like a CEO manages a team.
Automate the parts that used to eat your Fridays.
For more AI dev workflows like this: 👇
→ Go to substack.com/@humzakhalid
→ Subscribe for free (don't pay anything)
→ Get cheatsheets like this one every week
Which sprint do you think most devs skip? Drop it below. 🤯
♻️ Repost to save an engineer in your network 6 months of wasted effort.
Clayton Passos retweeted
RAGs vs Agents
Ask an LLM about your company's data and it will guess. The two patterns that fix this are RAG and agents, and they solve different problems.
RAGs: RAGs combine LLMs with retrieval to ground answers in 4 steps.
Step 1: The user query is embedded and sent to a retrieval step.
Step 2: Retrieval pulls the most relevant chunks from a knowledge base (PDFs, wikis, etc.)
Step 3: Those chunks are pasted into the prompt as context.
Step 4: The LLM writes the answer, grounded in the retrieved text.
One retrieval. One generation. Cheap, predictable, and easy to debug.
Agents: Agents wrap LLMs in a reasoning loop with tools to take action.
Step 1: The user query goes into the agent runtime. A reasoning loop wrapped around an LLM.
Step 2: The LLM reads the goal and picks a tool (Read, Write, Edit, Bash, etc.)
Step 3: The runtime executes the tool and feeds the result back to the LLM.
Step 4: The LLM reasons again, picks the next tool, and loops until the task is done.
More flexible. More tokens. Harder to debug because errors drift across steps.
The rule of thumb: Use RAG when the answer lives in your documents. Use an agent when the answer requires action on other systems.
Over to you: When do you prefer RAG over agent?
Técnicas para evitar duplicação dos dados:
- Desduplicação
- Imutabilidade
- Transational outbox pattern
- Idempotência
- Conciliação
- Configurações no próprio broker de mensagens (visibility timout)
