AI Agents
AI agents are LLM-powered systems that don't just answer — they plan steps, call tools and drive a task to completion: writing code, searching the web, booking, analyzing. 2026 became the year of agents, from coding agents to autonomous researchers. This page gathers all our coverage of agentic AI: launches, protocols (MCP, A2A), reliability and real-world use.
NEWThe DecoderGoogle lets third-party AI agents into Android Studio: Claude, Codex and 40+ more
Simon WillisonA Swarm of AI Agents Builds the Game Spawn, and Tokens Cost $3 Million a Month
TG: @ai_machinelearning_big_data (Machinelearning)Mistral Releases Large 4 Weights: Open 1T-Parameter Model for Coding
TG: @ai_machinelearning_big_data (Machinelearning)Google unveils Gemini 4 Carbon: 88% on vulnerability discovery and first place in the ranking
ElevenLabs BlogElevenLabs hands routine work to internal agents and saves 1,200 hours a month
GitLab BlogGitLab Orbit hits GA: agents need 4.5x fewer tokens
Habr AIDeveloper builds a RAG system on local models, boosting search accuracy from 20% to 53%
Habr AIOES Puts 30 AI Agents on Mining Haul Trucks and Lifts Haulage by 10%
ElevenLabs hands routine work to internal agents and saves 1,200 hours a month
ElevenLabs' internal ElevenAgents answer 5,000 questions a month, file Linear tickets and build customer demos. The main lesson: cleaning up the data schema matters…
GitLab Orbit hits GA: agents need 4.5x fewer tokens
GitLab launched Orbit, a context graph for the entire development lifecycle. In internal tests, agents used up to 4.5x fewer tokens and made up to 45x fewer retries.
Developer builds a RAG system on local models, boosting search accuracy from 20% to 53%
An engineer built a news-analysis AI agent without a single cloud API call, running everything on local models on his own hardware. A hybrid search approach fixed…
Skills for AI Agents Aren't Universal: Weak Models Need Their Own Version of the Prompt
AI agent on GPT-5.5 audits 2600 TravelLine autotests in two weeks
Study: A Single Random Failure Makes AI Agents Abandon a Working Tool 11.6% of the Time
Liquid AI releases LFM3: six multimodal 2B models with a 256K context window
AdaPRA: A Supervisor Agent Boosts AI Performance on Data Science Tasks by 24.5 Percentile Points
Researchers at Carnegie Mellon taught an AI agent to catch its own silent mistakes in data analysis using adaptive checklists with code verification — results on…
AOS-Bench tests AI agents: quality drops 13-23 points under parallel workloads
A new benchmark called AOS-Bench found that when agents built on Claude, GPT, and other models are handed several tasks at once instead of one, they lose up to 23%…
Wired Tested 6 AI Agents in a Day: Dots Leads, Gemini Spark Fails at Lunch
Anthropic reveals how it builds Claude with Claude: the brain separate from the hands
AI Agents Translated a 15,000-Page Proof Into 43 Million Lines of Code in 24 Days — and Found a 40-Year-Old Error
Anthropic Turns Claude Code Into a Platform: Claude Mods Flip the Agent Harness
Anthropic files S-1 for IPO at $2 trillion valuation: 85% margin, profitable Q2
Anthropic's S-1 discloses Q2 revenue of $11.6 billion (+1368%), operating profit of $974 million and an 85% gross margin. Ticker ANT, Nasdaq listing.
SLICE-Mem Cuts AI Agent Memory Token Costs by 92% Without Sacrificing Quality
Researchers at Hong Kong Polytechnic University have found a way to store AI agent memories without constant calls to a large language model — accuracy barely…