2025 Wrapped
Insights
- State of AI agents by Andrew Ng
- One year of agentic AI: six lessons from the people doing the work by McKinsey
- State of AI: an empirical 100 trillion token study with OpenRouter
- AI agents are winning hearts and wallets by G2
- State of AI report 2025 by Air Capital
- The state of AI agents by Lance Martin
- 8 learnings from 1 year of agents by PostHog
- The rise of computer use and agentic coworkers by a16z
- How small language models are key to scalable agentic AI by Nvidia
Stories
- Karpathy says it will take a decade before AI agents actually work
- Silicon Valley bets big on ‘environments’ to train AI agents
- AI agents still majorly struggle with real-world work
- Gartner: agents are at the top of the hype cycle
- Cloudflare & Browserbase: pioneering identity for AI agents
- Linux Foundation announces the formation of the Agentic AI Foundation (AAIF)
- AI agents aren’t ready for consumer-facing work — but they can excel at internal processes
- Silicon Valley builds Amazon and Gmail copycats to train AI agents
Commentary
- Real AI agents and real work by Ethan Mollick
- Why (senior) engineers struggle to build AI agents by Phil Schmid
- AI Agents have, so far, mostly been a dud by Gary Marcus
- Agent design is still hard by Armin Ronacher
- Aaron Levie and Steven Sinofsky on the AI-worker future by a16z
- Agents should be more opinionated by Vivek Trivedy
Learning
- Agentic AI course with Andrew Ng
- A comprehensive review of AI agents: transforming possibilities in technology and beyond
- AI agents from scratch
- You should write an agent
- Agents towards production
- Agents built from alloys
- Building an AI agent from scratch in Python
- 5-day AI agents intensive course with Google
- Lessons from building verticalized AI agents
- How to build an agent or: the emperor has no clothes
- Building more effective AI agents
- Context engineering for AI agents: lessons from building Manus
- Fundamentals of building autonomous LLM agents
- Prompting for agents
- AI agents in production: lessons from Rippling and LangChain
- How to train your agent: building reliable agents with RL
- Building the future of agents with Claude
- 3 ingredients for building reliable enterprise agents
- Building AI agents that actually automate knowledge work
- Stanford CS329A: self-improving AI agents
- Effective context engineering for AI agents
- Small language models are the future of agentic AI
- How to use Claude Code for everyday tasks
- Building coding agents with tool execution
- Simulations: the secret behind every great agent from Sierra
- Designing agentic loops
Research
- Measuring agents in production
- Agent-in-the-loop
- AgentFold: Long-Horizon Web Agents with Proactive Context Management
- When AIs judge AIs: the rise of agent-as-a-judge evaluation for LLMs
- Solving a million-step LLM task with zero errors
- The adoption and usage of AI agents: early evidence from Perplexity
- Building the web for agents
- Efficient agents: building effective agents while reducing cost
- Agent²: an agent-generates-agent framework for reinforcement learning automation
- Exploring autonomous agents: a closer look at why they fail when completing tasks
- The era of agentic organization: learning to organize with language models
- AI agents for economic research
- General social agents
- Virtual agent economies
- A framework for studying AI agent behavior: evidence from consumer choice experiments
- Training agents inside of scalable world models
Projects
- Parlant - natural language guidelines for agents
- Agent Lightning - seamless agent optimization
- Better Agents - standards for building agents
- Atomic Agents - lightweight and modular agent framework
- Airweave - context retrieval for agents
- Cognee - memory for agents
- Graphiti - knowledge graphs for agents
- AG-UI - Agent-User Interaction Protocol
- Acontext - agent context platform
- Upsonic - agent framework for fintech