Dee

Messi🐐🐐

[Patrick Collison @patrickc]

Assorted things I've been mulling over regarding agents in the economy:

• Companies are in some sense the original superintelligence. It's often the case that they can use their amortized cognitive surplus to advance their interests against those of certain consumers: relative to the optimization capacity of companies, consumers can be inattentive, underinformed, irrational, etc. (Think, say, of some of the classic insurance upsells at the end of a booking process, which is typically priced substantially above what a regular policy would cost.) Herbert Simon introduced the idea of bounded rationality, and firms often end up benefiting from it. Thanks to personal agents, we will probably get somewhat more rational in how we engage with companies.

• How exactly will outcomes change? Well, certain kinds of price discrimination will work less well (how do coupons work if every agent hunts fastidiously for them?), and certain kinds of customer cross-subsidy will change (today, people who forget to cancel subscriptions subsidize those who don't). Taken across the economy as a whole, I'm not sure what the incidence of all of this will be.

• In corporate strategy, people have long debated the relative importance of distribution and product quality. Perhaps you've made something better, but does anyone know about it? Won't people just continue to buy the ACME Corp Mousetraps? Personal agents will, I think, act as a kind of structural subsidy for product quality. How would markets look different if individuals always spent at least 10 hours researching their purchases? On the margin, focusing on making a product better is, I think, going to become a more effective strategy for companies. This is good!

• It true, a corollary is that information about product quality will become more important and impactful. Today, most people have very limited ability to aggregate and integrate such data, but agents will have insatiable thirst for such signals. (It's often said that data is becoming more valuable because it helps in the training of models. That's obviously true, but I think there's a secondary way in which it's more valuable simply as a complement to cheaper intelligence.)

• A big question is whether agents will increase superstar effects (rational agents all settle on the same product) or increase dispersion (because everyone has slightly different tastes and preferences, which agents are good at eliciting, finally overcoming the artificial clustering we've been subjected to).

• Many internet companies have monetized demand routing, which is in turn often predicated on short attention spans. "I'll click on the first thing I see." This has led to some amount of inefficiency, as companies try to propitiate capricious ranking algorithms and incur various placement fees. It's not all deadweight loss (willingness to pay is itself a kind of signal, which is valuable to buyers), but it's not great. How will the nature and value of this routing change in a world of personal agents? I don't know the answer, but I think that the effects might be pretty big.

• Much of what I'm writing hinges on the assumption that personal agents will be on the consumer's side. I suspect that that will be the winning strategy, but whether it's true, and what it means to be true, involves a lot of other questions.

Taken as a whole, the market is a massively parallelized, decentralized, and highly imperfect value attribution machine, and, taking stock of all of the effects, it seems to me that the quality point may well be the biggest: that agents will change the nature of the backpropagation such that the rewards for better products increase. Maybe this is too optimistic, but I suspect (and hope) that companies that make things that they're proud of will feel that the universe is a little more partisan in their favor.

[Vishal Singh 🥑 @vishalsingh2972]

Do yourself a favor if you're serious about AI Engineering:

– Learn LLM APIs, tool calling, structured outputs and context engineering

– Find one boring, repetitive task people genuinely hate doing

– Check if simple automation can solve it before building an agent

– Define what success looks like before writing code

– Build one useful agent, not a team of 10 agents talking to each other

– Give it the right tools, context and memory only when needed

– Learn MCP, but use it when it actually makes integrations easier

– Build a small eval dataset from real examples before obsessing over prompts

– Test messy inputs, edge cases and failure scenarios, not just happy paths

– Validate tool inputs and outputs with strict schemas

– Make the agent verify its work instead of confidently guessing

– Teach it when to ask questions, stop or hand control back to a human

– Give tools minimum permissions and keep user data isolated

– Keep API keys and sensitive data out of prompts and logs

– Treat websites, documents and tool responses as untrusted input

– Sandbox risky execution and require approval before sending, spending or deleting

– Add timeouts, limited retries and safeguards against duplicate actions

– Save progress so long-running tasks can recover from failures

– Set limits on tokens, execution time and cost per task

– Use smaller models for easy tasks and reasoning models when needed, based on eval results

– Cache safe, repeatable work without serving stale or private data

– Trace model calls, tool usage, errors and decisions so you can debug failures

– Evaluate both the final answer and the steps the agent took

– Track task success rate, latency and cost per completed task

– Run evals in CI and block changes that break existing behavior

– Test new models and prompts on a small percentage of traffic before rolling them out

– Add fallbacks and a rollback plan for when models or tools fail

– Turn every real user complaint into a new test case

– Show users progress, clear errors and what the agent actually completed

��� Deploy it with a live link people can genuinely try

– Write a simple README explaining how it works and its limitations

– Record a 60-second demo showing a real problem getting solved

– Open-source the code and put the results on your resume

One real project will teach you more than 100 AI tutorials.

And give you something better than a certificate: proof you can build.

📌 Bookmark this + send it to someone prepping for AI interviews.

[AI_Explorer @ai_explorer25]

Best 15 accounts to follow in AI:

@karpathy = LLMs king
@steipete = built openclaw
@gregisenberg = startup ideas king
@rileybrown= vibecode king
@jackfriks = solo apps king
@levelsio= startups king
@marclou = startups king
@EXM7777 = AI ops + systems king
@eptwts = AI money twitter king
@ai_explorer25= AI queen
@godofprompt =prompt king
@vasuman = AI agents king
@AmirMushich = AI ads king
@0xROAS = AI UGCs king
@egeberkina = AI images king

Follow them all and learn.

[HALO | CAREER SPECIALIST @Halosznn_]

IF YOU ARE 21, 22, 23, 24, 25, 26, 27, 28 or 29 years old

Here are Internships, scholarships & grants you need to apply for this October:

[Tech with Mak @techNmak]

There are 2 career paths in AI right now:

The API Caller:
Knows how to build with LLMs.

The Architect:
Knows how LLM systems are built.

If you want to move toward the second, Stanford has one of the best free LLM engineering playlists on YouTube:

CS336: Language Modeling from Scratch.

The 2026 course has 18 lectures covering almost the entire LLM stack -
➡️ Build the model: Tokenization, Transformers, architectures, MoE
➡️ Understand the hardware: FLOPs, memory, GPUs, TPUs
➡️ Make it fast: Triton, GPU kernels, parallelism, distributed training
➡️ Train it: Scaling laws, data collection, filtering, deduplication
➡️ Run it: Inference, evaluation
➡️ Post-train it: SFT, RLHF, RLVR

Plus multimodality.

And you don’t only watch lectures.
> You implement the tokenizer, Transformer and optimizer.
> You write FlashAttention2 in Triton.
> You build memory-efficient distributed training.
> You turn raw Common Crawl dumps into pretraining data.
> You fit a scaling law.
> You use SFT + reinforcement learning to train a language model for mathematical reasoning.

Stanford says students write at least an order of magnitude more code than in most other AI classes.
Stanford CS336. Spring 2026. 18 lectures. Free on YouTube.

Choose your path.

(Playlist in the comments)

♻️ Repost to save someone $$$ and a lot of confusion.

[Kyrox @kyr0stack]

Andrej Karpathy:

"Prompting is going away.

Delete everything, keep Graph."

In 1 hour he shows how to build Graphs, and why it's the only thing that will be left standing at the end

LLMs → Prompts → Agents → Graphs

a prompt gets you one answer and then forgets the job

an agent finishes the job and then forgets what it learned

a graph keeps every finished job and turns it into the starting point of the next one

the first 10 minutes alone are worth more than most paid courses

watch the lecture first, then save the full guide on Graphs below ↓

[fleyta @fleyta88]

this is certified f**king gold

Andrej Karpathy just showed a cleaner way to build research agents, and it kills the dumbest loop in AI:

search → answer → forget → search again

he basically turns memory into a real system:

raw sources stay raw
the agent reads them once
turns the useful parts into linked wiki pages
then checks the wiki first every time a new question comes in

so the loop changes:

> question comes in
> wiki gets searched first
> gaps send the agent back to source material
> the answer gets written back into the wiki
> the system leaves the next run smarter than the last one

that’s the trick

the “memory” is not buried in some giant chat transcript anymore

it lives in files the agent can actually work with:
> pages it can read
> links it can follow
> notes it can update
> gaps it can spot
> stale claims it can fix

after enough runs, you stop getting one-off answers

you get an actual knowledge base

that’s why this feels bigger than just “better RAG”

a normal research agent can handle 100 questions and leave behind 100 dead chats

Karpathy’s setup can handle the same 100 questions and leave behind a research system that compounds

the architecture is stupidly simple:

> raw/ = source of truth
> wiki/ = maintained knowledge
> schema = rules for how the whole thing gets updated

search still matters

it just stops being the whole product

that’s the part I’d steal immediately:

don’t make the agent remember harder

make it write something worth remembering

[Avi Chawla @_avichawla]

A full-stack LLM engineering roadmap covering:

> Prompt engineering
> RAG systems
> Fine-tuning
> Agents
> LLM deployment
> LLM optimization
> Safety, evals & observability
> Context engineering

Free and open-source resources in the article below.

(bookmark this)

[Suraj Sharma @suraj_sharma14]

If you learn AI Infrastructure Engineering, you will never be unemployed again.

Stage 1: Systems & GPU Foundations
- Learn: C++, Rust, CUDA memory hierarchy, PCIe vs NVLink, thread blocks and warp scheduling.

- Practice: Write a custom CUDA kernel for matrix multiplication from scratch and profile it against cuBLAS.

- Why: You cannot optimize what you do not understand at the silicon level.

Stage 2: Transformer Inference Physics
- Learn: Prefill vs Decode phases, arithmetic intensity, memory bandwidth bottlenecks, KV-cache math.

- Practice: Profile a HuggingFace model with Nsight Systems to identify the exact memory bottleneck during token generation.

- Why: LLM inference is almost always memory-bound, not compute-bound.

Stage 3: Modern Serving Engines & Batching
- Learn: vLLM, SGLang, PagedAttention, continuous batching, chunked prefill.

- Practice: Deploy a 70B model and tune chunk sizes to maximize throughput without starving decode requests during long-context prefills.

- Why: Naive batching leaves 60% of GPU VRAM wasted. PagedAttention fixes this.

Stage 4: KV-Cache & Memory Optimization
- Learn: Prefix caching, KV quantization, CPU offloading, multi-turn cache reuse.

- Practice: Build a routing proxy that directs requests with identical system prompts to the same replica to share KV blocks.

- Why: Reused cache is free speed. It can slash Time-To-First-Token (TTFT) by 80%.

Stage 5: Quantization & Compression
- Learn: FP8, INT4, AWQ, GPTQ, Sparsity, TensorRT-LLM calibration.

- Practice: Serve a model in FP8 vs FP16 and benchmark the exact perplexity drop against the latency and VRAM gains.

- Why: Quantization is the only way to fit frontier models on edge GPUs and protect your margins.

Stage 6: Kernel-Level Engineering
- Learn: Triton, FlashAttention, CUDA graphs, operator fusion.

- Practice: Write a fused Triton kernel for RMSNorm or Softmax and benchmark it against the native PyTorch implementation.

- Why: Python overhead kills inference. Fused kernels save milliseconds that compound at scale.

Stage 7: Distributed Inference & Parallelism
- Learn: Tensor parallelism, Pipeline parallelism, Expert parallelism (for MoE models).

- Practice: Shard a 405B parameter model across 8 nodes and measure the communication overhead between GPUs.

- Why: Single-GPU inference is dead for frontier models. You must master the shard.

Stage 8: Speculative Decoding
- Learn: Draft-target models, Medusa heads, acceptance rates, n-gram drafting.

- Practice: Build a pipeline where a local 3B model drafts tokens for a cloud 70B model to verify in parallel.

- Why: 2x decode speed at zero quality cost is the closest thing to a free lunch in inference.

Stage 9: Multi-Node & Hardware Interconnects
- Learn: NCCL, RDMA, InfiniBand, NVLink, Disaggregated Prefill/Decode architectures.

- Practice: Set up a multi-node cluster and profile the network latency of tensor parallelism across nodes vs within a single node.

- Why: Network latency is the new GPU bottleneck. Disaggregating prefill and decode is the 2026 meta.

Stage 10: Cluster Orchestration & GPU Scheduling
- Learn: Kubernetes GPU operators, Ray, Slurm, MIG (Multi-Instance GPU) partitioning, KEDA.

- Practice: Build a queue-based autoscaler that spins up spot GPUs based on pending inference requests and drains them when empty.

- Why: Idle H100s burn $3+/hr. FinOps and scheduling are now core infra responsibilities.

Stage 11: AI Gateways, Routing & Observability
- Learn: TTFT/ITL SLOs, semantic routing, DCGM metrics, OpenTelemetry for LLMs.

- Practice: Build a gateway that routes simple queries to a quantized local model and complex reasoning to a frontier API based on prompt complexity.

- Why: Routing protects your margins and DCGM metrics tell you when your GPUs are silently throttling.

Stage 12: Public Benchmarks & Teardowns
- Learn: Reproducible methodology, latency/throughput Pareto curves, cost-per-token analysis.

- Practice: Publish a teardown comparing vLLM vs SGLang vs TensorRT-LLM on your specific hardware with full configs.

- Why: Public proof of hardware mastery gets you hired instantly by top AI labs.

Wrappers are a commodity. AI Infrastructure is the physics of scale.

The modern AI Infra Engineer builds the silicon nervous system for global intelligence.

Bookmark and Repost!

[freeCodeCamp.org @freeCodeCamp]

LLMs are pretty impressive tools. But on their own, they can't "remember" anything.

That's where vector stores come in and enhance their capabilities.

In this article, @manishmshiva explains what vector stores are and how they work, teaches you about embeddings, and more.

https://t.co/lnwmZCk4NP

[Rohit Ghumare @ghumare64]

As an AI Engineer. Please learn:

> Harness engineering, not just prompt engineering
> Context engineering, not just long prompts
> Prompt caching vs. semantic caching tradeoffs
> KV cache management, eviction, reuse, and memory pressure at scale
> Prefill vs. decode latency and why they optimize differently
> Continuous batching, paged attention, and throughput optimization
> Speculative decoding vs. quantization vs. distillation tradeoffs
> INT8, INT4, FP8, AWQ, GPTQ, and when quantization hurts quality
> Structured output failures, schema validation, repair loops, and fallback chains
> Function calling reliability, tool contracts, argument validation, and idempotency
> Agent guardrails, loop budgets, tool budgets, and termination conditions
> Model routing, graceful fallback logic, and degraded-mode UX
> RAG architecture: chunking, embeddings, hybrid search, reranking, and freshness
> Retrieval evals: recall, precision, grounding, attribution, and citation quality
> Evals: golden sets, regression tests, adversarial tests, LLM-as-judge, and human evals
> LLM observability as a first-class discipline: traces, spans, tokens, latency, errors, and drift
> Cost attribution per feature, workflow, tenant, and user journey, not just per model
> Safety engineering: prompt injection defense, data leakage prevention, and permission boundaries
> Multi-tenant isolation, cache safety, and cross-user context contamination prevention
> Fine-tuning vs. in-context learning vs. RAG vs. distillation, and when each is the wrong tool
> Latency, quality, cost, and reliability tradeoffs across the full inference stack
> Production failure modes: hallucinated tool calls, malformed JSON, stale retrieval, runaway agents, and silent eval regressions
> Shipping LLM systems as reliable infrastructure, not demos wrapped around prompts

https://t.co/D2n8y3Ktnv

[Mark Ajzenstadt @mardehaym]

Everyone is talking about AI.

AI, AI, AI.

Nobody wants to talk about the unsexy part: the data foundation.

Here are the steps that make any AI that comes next actually work for the enterprise:

[Tanzil Chowdhury @iamtanzil_]

Every serious vibe coder should have bookmarked it.

25 vibe coding resources to keep yourself ahead of the competition.

1. https://t.co/CgnzwKME3r — design prompt library with 1200+ AI-ready design prompts to help you design beautiful websites fast with AI.
2. Minimal Gallery (https://t.co/J1fPSn1KpA) — curated high-end sites to feed your agent
3. Kage (https://t.co/IHOcieuxPv) — real UI inspiration mapped straight to prompts
4. Refero Styles (https://t.co/xrnqcp6NRp) — 2,000+ real product styles with typography
5. Component Gallery (https://t.co/Xo7ZU8P5tA) — 2,600+ examples of how top design systems solve the same element
6. AppShot Gallery (https://t.co/Z0ILApmLon) — real app screenshots for mobile
7. DESIGNmd (https://t.co/4Hn2Mjrh29) — design systems as markdown your agent can read
8. VibePrompt (https://t.co/xkYYIewqtw) — ready-made prompts for dashboards and landing pages
9. https://t.co/H3YmQX68Ph (https://t.co/H3YmQX68Ph) — component registry that plugs into agents via MCP
10. Kinetics (https://t.co/XgNvWvYhop) — 150+ motion effects with React code and prompts
11. shadcn/ui (https://t.co/pSXS4kZiE9) — still the gold standard for clean, reusable UI
12. Aceternity UI (https://t.co/RhoC6OX8eH) — 200+ animated React/Tailwind components
13. Magic UI (https://t.co/heZ00fkLry) — beautifully designed components for modern web apps
14. Motion Primitives (https://t.co/criRCDcX8n) — ready-to-use animation and interaction components
15. Uiverse (https://t.co/Negddg28wS) — community-built UI elements and interactive components
16. UIAble (https://t.co/GWbsNpbdZz) — a collection of ready-to-use UI components
17. mapcn (https://t.co/tV4opEZSFL) — map components, markers, routes, popups
18. MicroKit UI (https://t.co/L27FjbMwSd) — small micro-interactions that make interfaces feel polished
19. Liquid Glass (https://t.co/iFjZs9zCsW) — glass refraction components and effects
20. CSS Text Effects (https://t.co/GZkLIsobRf) — creative CSS text animations and effects
21. Circle Loaders (https://t.co/I36Cydh1zU) — creative loading animations and loaders
22. Gradient Buttons (https://t.co/WUNXnN8OLa) — gradient button styles and effects
23. Kitbitz (https://t.co/9LcTjAcI6V) — 2,000+ hand-drawn illustrations
24. 3Dicons (https://t.co/qaAFF77ntp) — a library of free 3D icons for modern interfaces
25. Anime.js (https://t.co/VgOPVGH4oY) — lightweight JavaScript animation library for the web

If you're building with AI, this list will save you hours.

Save it. 🔖

[Marco @maarcoofdezz]

Anthropic contrató a este ingeniero a $250K-$750K al año porque sabe cómo construir arneses para sistemas multiagente

En este taller de 15 minutos, muestra exactamente cómo construir uno desde cero

IA → Agentes → Arnés → Bucles → Gráficos

paso 1 → comienza con el SDK de Claude Agent - el arnés maneja bucles, contexto y sandboxing

paso 2 → separa el cerebro de las manos - razonamiento en un lugar, herramientas en un sandbox, 60% más rápido al primer token

paso 3 → ejecútalo del lado del servidor y registra cada paso - cierra tu laptop y sigue ejecutándose, los fallos se reanudan desde el registro

paso 4 → haz que el fallo sea barato - reintenta sandboxes muertos y reproduce contexto perdido en lugar de empezar de nuevo

paso 5 → convierte los registros de ayer en nueva memoria y habilidades - el arnés despierta más inteligente

Anthropic llama a esto "soñar"

La mayoría de las personas pasa semanas construyendo esto desde cero.

No tienes por qué hacerlo.

Guárdalo, cuando llegue el momento de construir agentes que realmente funcionen en producción, este arnés puede ahorrarte semanas de trabajo

[Tobias van Schneider @vanschneider]

There is *a lot* of software that hasn’t been built across hundreds of industries because our best paid designers and engineers were occupied with building social networks and ad networks.

"The best minds of my generation are thinking about how to make people click ads." — Jeff Hammerbacher

Meanwhile, huge parts of the physical economy were left running on spreadsheets, email, paper forms, ancient desktop software and enterprise systems installed sometime around 1999. They still use it today.

Construction. Manufacturing. Logistics. Agriculture. Waste management. Energy. Shipping. Maintenance. Laboratories. Local government. Field services. Industrial equipment. Healthcare administration. The list is endless.

A lot of these places were simply too small, too "unsexy" or too expensive to serve with traditional software economics, especially after engineers became one of the best paid career choice.

But small teams can now build specialized tools for markets that previously couldn’t justify the best of the best engineering talent.

And even further, lets not forget about all the terrible enterprise software out there owned by massive corporations who locked in clients since the late 90s and didn't change a single thing since then. All these companies are completely vendor locked due to switching cost, but that switching cost is slowly disappearing.

Maybe I am too optimistic here, but there are entire industries still waiting for their first piece of software that is truly designed for the people using it.

The opportunities right now are endless.

[JezebelReborn🧙‍♀️ @JezebelReborn]

EXPOSED: “Olu of Warri, Tompolo and Osahon Idahosa collect ₦2.7tn annually from the Nigerian Federal Government through pipeline contracts.”

Asari Dokubo said Tompolo is involved in the contracts but receives far less than the figures being alleged, while he himself has no involvement in the pipeline contracts.

He also claimed Arthur Eze previously held the contract and that he saw an Orji Uzor Kalu ( former Gov of abia state) -owned vessel operating in their waters.

— Asari Dokubo blows hot

[HALO | CAREER SPECIALIST @Halosznn_]

Goldmineee💗

Graphic Design (Beginner - Advanced Level)
https://t.co/DQ7zgTtzQ8

Full stack Development
https://t.co/Hh39Rz88DZ

https://t.co/NFIUsY6CaG

Project Management
https://t.co/LrAOw7bmVM

Vibecoding
https://t.co/ilXznaqOVp

Digital Marketing
https://t.co/FUTpiJVSEd

Video Editing - Adobe Premiere Pro
https://t.co/TXaoJQMB9U

Copywriting, SEO, Content writing
https://t.co/z5kPnAr8CK

Advanced CapCut course
https://t.co/aAUdLnLWl0

Free

Video Editing with AI
https://t.co/WIBye9QV0G

Frontend Development
https://t.co/MUlKY8kI90

Get free AI updates, resources, opportunities
https://t.co/ZFPhuDOaD0

UI/UX Full course
https://t.co/cEygd3JwaX

[Careerspark by kay || Digital baddie @careerspeakshq]

If your are interested in AI, marketing and automation, including LLM APIs, Python/Node, Zapier/Make/n8n and prompt engineering.
There is an internship opppurtunity

[Dhanush N @Dhanush_Nehru]

AI security is about to be the highest-paid skill in tech.

Everyone's building AI agents. Almost nobody's securing them.

Master these 10 concepts and you'll be in the top 1% of AI security engineers. 🧵

[Peter O. @oluwagbemigaptr]

Join the Sahara Energy Talent Pool

They welcome applications from professionals across a wide range of disciplines, including but not limited to:

Corporate Communications, Branding, Advertising, Public Relations

Marketing, Business Development, Sales and Client Relations

Government Relations, Community Relations, Corporate Social Responsibility

Supply Chain, Purchasing, Procurement, Inventory Management and Demand Planning

Fleet Management, Transport and Logistics

Security and Health, Safety, Security and Environment

Facilities, Office Management, Real Estate, Construction

Strategy, Business Performance

Accounts, Finance, Risk, Audit, Internal Control

Human Resources, and General Administration

Legal, Corporate Governance and Company Secretariat

Drilling Operations, Well Services

Engineering, Oil and Gas Operations, Instrumentation

Geology, Geoscience, Sub Surface Engineering

Information Technology, Artificial Intelligence, Robotics, Software and Hardware

Sustainability (Environment, Social and Governance)

Submit your CV today and discover where your potential can take you.

https://t.co/VAodaFRvPq

Retweet to your network.