Knowledge catalogue
// The Consistency Illusion // Multi-agent debate can make agents agree on the final answer while their underlying reasoning stays misaligned. This work finds that consensus on the output hides disagr
Great tips. In practice, this is how it roughly looks to run agents autonomously for hours or days. /goal or /loop to keep it going. Verification is crucial here. Seeing a number of benchmarks showing
New paper on how AI agents are reshaping knowledge work. This is a nice economic read on where agents actually change knowledge work to meet that gap directly. (bookmark it) It studies agent adoption
The point is that you should start implementing ways to encode instructions/prompts with clear goals inside automations. Nothing new but newer LLMs are being trained to perform for longer duration uni
Great paper on self-improving agents: This was one of the standout AI papers of the week. (bookmark it) It tackles a question most self-improving AI agents ignore: is the agent actually discovering an
Super-powerful AI models will launch in the coming weeks. We are looking at a potential step change in model capabilities. The biggest mistake right now is to lock into one vendor. I say this not only
The Top AI Papers of the Week (May 31 - June 7) - LEAP - AutoLab - Learn From Your Own Latents - Reusable Context Engineering - Self-Revising Discovery Systems - Scaling Laws for Agent Harnesses - Dis
This was one of the standout AI papers of the week. (bookmark it) It tackles a question most self-improving AI agents ignore: is the agent actually discovering anything, or just remixing what it alrea
// Continual Learning Bench // One of the research areas with lots of investments is continual learning. While there are many efforts, there is very little progress in measuring it. So the big questio
New research from Renmin University. Treat skill selection as a harness in its own right. If you design skill routing for personal or edge agents, this work argues that the selection layer is a first-
// Agents' Last Exam // Agents' Last Exam is a living benchmark of over 1,000 economically valuable tasks, built with 250+ industry experts and mapped to the U.S. federal occupational taxonomy. The ha
Find an important unsolved problem you care about. Then use AI to solve it. Go deep! Talk to people. Build a community. It might take you months or years, but always know that AI capabilities will onl
// The Meta-Agent Challenge // How good are current agents at self-improving? This is a great paper covering some of the challenges. They propose the Meta-Agent Challenge (MAC), where they give a codi
I am hooked on Dynamic Workflows! The idea of generating harnesses on the fly is so compelling that I reverse-engineered it for my agent orchestrator. And then I built a monitoring dashboard (as an HT
NEW: NVIDIA ships 550B MoE open model for long-running agents. Very exciting times to see more open models to support local long-running coding agents. Today we're shipping Nemotron 3 Ultra. A 550B Mo
Outstanding paper on long-horizon agents. (bookmark it) Similar to humans, how do you make agents persist on a difficult task, and how is that useful? And which models today work well on this? This ne
The post announces that the author will share insights about their use of dynamic workflows and explain their reasoning behind building a custom implementation. This likely covers practical applicatio
Another banger open-source release. Miso One is an 8B text-to-speech model with real emotional range, so voiceovers carry warmth, hesitation, and excitement instead of sounding flat. It's purpose-buil
New research from Google. Just shows the impressive results you can get from custom agent harnesses. LEAP wraps a general-purpose LLM in an agentic scaffold that grounds every step in the Lean compile
Nice primer on post-training reasoning data. (bookmark it) This is one of the first primers to pull the scattered post-training reasoning-data literature into one place, synthesizing over 150 public s
This SkillOpt paper from Microsoft is a must-read! (bookmark it) I was a bit skeptical of the results reported in the paper when I shared it a few days ago. However, I managed to integrate it into my
// Scaling Behavior of Single LLM-Driven Multi-Agent Systems // Does adding more agents actually make a multi-agent system better? It's possible that collective intelligence emerges from interaction d
// State-Externalizing Harnesses // A new paradigm is emerging on how to effectively build agents and harnesses. If there is a state that the environment can maintain reliably, it probably doesn't bel
MiniMax M3, an upcoming AI model, is expected to be released soon and will undergo comprehensive testing within a custom coding agent framework. A detailed technical review of the model's performance
// Reusable Context Engineering // Context bloat quietly kills long-horizon runs, but you can fix it from the outside without fine-tuning the underlying agent. (bookmark this) Context management is us
Very good advice on self-improving agents. (bookmark it) This is something I am seeing in my own experiments with coding agents and harnesses for long-horizon tasks. What I have found is that stronger
PewDiePie has developed and released his own AI agent orchestrator, representing a trend toward independent creators building custom AI infrastructure rather than relying on existing platforms. The po
// The Efficiency Frontier // Cool paper on context management. As agents reuse the same documents and histories across many turns, the cheapest context strategy is not fixed. This work describes a pr
The Top AI Papers of the Week (May 24 - May 31) - SkillOpt - AutoScientists - The Efficiency Frontier - Language Models Need Sleep - Adapting the Interface, Not the Model - Forecasting Scientific Prog
I did a talk on LLM Wikis and HTML artifacts recently, if you are curious to learn more on the topic: https://academy.dair.ai/events/cmovobp97000904l5h0n9a2yz Doing a second session and a few releases
In a few months, people will start to realize how fundamentally important MCP for agents is. It's not even about connecting tools. There are many ways to do that. It's about the types of abstraction i
Increasingly, HTML Artifacts are becoming a core part of how I work with AI agents. Long-horizon agent sessions need a better way to surface insights about what work it has done. This may not be obvio
The efficiency frontier! Where do you think GPT-5.6 will land? Claude Opus 4.8 has landed on DeepSWE Bench, posting a 58% Pass@1 and taking #2 overall behind GPT-5.5. It continues a broader trend: sli
// Scaling Laws for Agent Harnesses // If you build agent harnesses, this one is worth your time. (bookmark it) Most harness tuning treats every token and tool call as if volume is all that counts. Ne
Banger paper from Harvard. AutoScientists drops the central planner entirely. Agents interpret shared experimental data, self-organize around promising directions, evaluate proposals before resource a
For future-proof, build AI that's composable. Regardless of what you use, all these should be composable, iterative, and customizable: - LLMs - Evals - Automations - MCP/CLI tools - Skills/Memory/Cont
It's crazy that this is even possible today. It inspired me to build my own self-improving coding agent with simple read, write, bash,... I already used the coding agent to build an entire production-
Stronger models do not always need lighter harnesses. Everyone believes more structured harnesses universally improve reliability, and that higher-capability models need proportionally less structural
// Your Agents are Aging Too // Huh!? They need 'sleep,' and now they are aging? Joke aside, great write-up on reliable agentic engineering. This new research introduces AgingBench, a longitudinal rel
Just built an insane new agent skill. It can perfectly extract slides from YT videos, then write notes, images, transcripts, and slides into Obsidian vaults. An HTML artifact allows me to navigate and
// Language Models Need Sleep // Let your agents 'sleep', folks. On a serious note, this is a fascinating paper on getting the most from long-horizon agents. Here is the problem with agents today: Att
New agent skill to convert YouTube videos to slides and notes. Just built an insane new agent skill. It can perfectly extract slides from YT videos, then write notes, images, transcripts, and slides i
System scaling is the next real bottleneck in agentic AI. If you build agent orchestration layers, this is a clean map of where the engineering leverage actually sits. The labs own the model. You own
/goal is really insane! It's how you can get the most out of coding agents today. For efficiency, I find it works best when you do planning before /goal. This ensures the agent has the right context a
New paper from Microsoft on Self-Evolving Agent Skills New research from Microsoft Research I see a lot of AI engineers handwriting agent skill docs and hope they generalize. Probably not optimal. Thi
// The Efficiency Frontier in LLMs // (bookmark this one) How much are you overpaying for context you do not need? It turns out that context costs dominate production LLM bills, and the right strategy
The Top AI Papers of the Week (May 18 - 24): - AIRA - MetaCogAgent - Memory as a Model - Code as Agent Harness - Weak-Model Critic-Comparator - OpenAI Disproves the Unit Distance Conjecture - Producti
This post emphasizes the growing demand for engineers and researchers in advancing technology fields, advising people to focus on mastering fundamental concepts rather than being distracted by hype or
// Adapt the Interface, Not the Model // I am fascinated by the results across my cheap-model-plus-good-harness builds. This new paper also shows good signs of the code-as-agent-harness thesis. The id
Can frontier models forecast scientific progress? Mostly no, but here is why. This work looks at 4,760 scientific events across disciplines. Frontier models can identify plausible research directions
Learn anything with our new /lesson-generator skill Just released my new /lesson-generator skill. Use it with your agent to learn anything: - generate lessons/courses on any topic - include nano-banan
NEW paper worth reading. A full agentic workflow can be distilled into model weights and run at roughly 100x lower inference cost while preserving near-frontier task quality. The workflow includes mul
The best way to learn AI is to build with agents. To help with that, we've launched hands-on labs and a new series on Agentic Engineering. First topic: Agent Skills. Next in the pipeline: planning, co
Learn how to build LLM Wikis and LLM Artifacts. New VIDEO: From LLM Wikis to LLM Artifacts Shared all my thoughts on why LLM wikis and HTML artifacts are a big deal. Plus, new tools to help you build
New VIDEO: From LLM Wikis to LLM Artifacts Shared all my thoughts on why LLM wikis and HTML artifacts are a big deal. Plus, new tools to help you build wikis and artifacts with agents. Just getting st
I highly recommend this. The Agentic Review is a new podcast from @QodoAI hosted by Itamar Friedman and Nnenna Ndukwe, and it's a great AI coding show that's neither hype nor doom. It's honest convers
If you design production agent systems, this matters. Most devs accidentally let their framework defaults make critical architecture decisions without thinking it through. This paper shows you how to
// Memory as a Model // The paper augments any LLM with a separate trained memory model that stores, retrieves, and integrates facts on its behalf. It decouples memory updates from base-model weight u
Self-improving AI is a big deal! As a first step, I've been exploring how much of the post-training can be automated. Here is a first post on how I am using @FireworksAI_HQ Agent to automate LLM fine-
Very interesting results from this NanoGPT-Bench eval. There is so much talk about self-improving agents. But can coding agents do real AI R&D? @IntologyAI reports that Codex, Claude Code, and Autores