Author · editorial director
François Chollet
François Chollet, AI researcher, creator of Keras and of the ARC-AGI benchmark, co-founder of Ndea and of the ARC Prize. Formerly at Google (2015-2024), author of “Deep Learning with Python”, named to the TIME100 AI in 2024.
His role on the site
François Chollet is the editorial director of Info Agent IA: he sets the line (measure rather than announce, date and source every figure) and the method used to assess agents, tools, MCP servers and models. Pages are produced from automatically collected public data, following that method, and then reviewed. Read the methodology.
Career
- 2012 Engineering degree (MEng) from ENSTA Paris. (source : Wikipedia)
- 2015 Releases Keras, an open-source deep learning library, then joins Google. (source : GitHub keras-team/keras)
- 2017 Publishes “Xception: Deep Learning with Depthwise Separable Convolutions” (CVPR) and the 1st edition of “Deep Learning with Python” (Manning). (source : CVPR 2017)
- 2019 Publishes “On the Measure of Intelligence”, which proposes a definition of intelligence based on skill-acquisition efficiency and introduces the ARC benchmark (Abstraction and Reasoning Corpus). (source : arXiv 1911.01547)
- 2021 2nd edition of “Deep Learning with Python”; receives the Global Swiss AI Award in December. (source : Manning)
- 2024 Launches the ARC Prize with Mike Knoop, a one-million-dollar competition to solve ARC-AGI. (source : arcprize.org)
- 2024 Named by TIME among the 100 most influential people in AI (TIME100 AI, September). (source : TIME)
- 2024 Leaves Google in November, after more than nine years. (source : Google Developers Blog)
- 2025 Co-founds Ndea with Mike Knoop, a research lab working on general AI through program synthesis; the ARC Prize becomes a non-profit foundation. (source : ndea.com)
Publications and projects
- On the Measure of Intelligence · arXiv, 2019
- Deep Learning with Python (2nd ed.) · Manning, 2021
- Xception: Deep Learning with Depthwise Separable Convolutions · CVPR, 2017
- Keras · GitHub, 2015
His content on the site
- Measuring progress toward AGI: benchmarks, limits, saturation, and contamination
- When will AGI arrive? Published forecasts, their methods, and why they differ
- The history of AGI: from Turing to today's labs
- AGI risks: what researchers say, without alarmism
- AI Act: a compliance checklist for businesses using AI agents and tools
- Getting started with an AI agent: choosing tools, setup, first tasks, and safeguards
- Writing AI agent instructions: system prompts, AGENTS.md and CLAUDE.md files, and examples
- Connect an AI agent to Telegram, WhatsApp, Slack, or Discord: methods and risks
- AI agent skills: how they work, how to install them, and how to build and verify them
- Deploying AI agents in business: use cases, selection, pilots, governance, and measurement
- Deep research with AI: tools, methods, and source verification
- ARC-AGI: the test, its versions, and published scores
- ARC Prize: the competition, its rules, editions, results, and lessons
- Choosing a language model: criteria, benchmarks, pricing, context, and licensing
- Choosing an MCP server: trust, maintenance, permissions, and local or remote hosting
- Claude Code reference guide: installation, subscriptions and API access, commands, CLAUDE.md, MCP, subagents, and best practices
- Codex CLI reference guide: installation, approvals, sandboxing, AGENTS.md, and costs
- Writing your company’s AI use policy: what to include, sample rules, and rollout
- Computer use agents: how they work, available tools, reliability, and security
- CrewAI reference guide: agent teams, tasks, tools, flows, and deployment
- Cursor reference guide: the AI code editor, modes, rules, pricing, and limits
- The limits of deep learning: what models struggle with, according to François Chollet and other researchers
- Evaluating an AI agent: test sets, metrics, public benchmarks, and production monitoring
- Fine-tuning vs. RAG: how to choose, costs, data requirements, and examples
- Generalization at the heart of intelligence: François Chollet's framework, examples, and measurement
- Quantization and GGUF: how they work, Q4 and Q8 levels, quality trade-offs, and choosing a file
- Hermes Agent reference guide: installation, memory and skills, models, costs, and limitations
- Hosting an AI agent on a VPS: server selection, installation, security, backups, and costs
- LangGraph reference guide: state graphs, memory, tools, and deployment
- How We Get To AGI: François Chollet’s talk, chapter by chapter
- Getting started with an LLM API: keys, your first request, settings, costs, and best practices
- Understanding LLM API pricing: tokens, input and output, caching, batch processing, and budgeting
- Local LLM reference guide: software compared, models by machine size, performance, and uses
- AI automation in Make and Zapier: features, agents, limits, and pricing
- Monitoring an AI agent in production: logs, traces, alerts, and observability tools
- Multi-agent systems: when several agents work better than one, architectures, costs, and pitfalls
- n8n and AI: A complete guide to nodes, agents, memory, hosting, and costs
- OpenClaw reference guide: installation, configuration, models, skills, messaging, security, and costs
- Program synthesis: principles, history, and a possible path to AGI
- RAG reference guide: how it works, chunking, embeddings, vector databases, evaluation, and common mistakes
- Reduce AI agent costs: model selection, caching, context, routing, and budgets
- Securing an AI agent: permissions, secrets, sandboxing, human approval, and logging
- Reasoning models: how they work, what they improve, and measured limits
- Test-time adaptation: how it works and what results show
- Working effectively with a coding agent: scope, review, tests, and costs
- Voice agents: how they work, platforms, latency, costs, and use cases
- Mistral AI: models, pricing and agents
- Open source or proprietary AI agent: how to decide
- AI agent vs. chatbot: what's the difference?
- Build your own MCP server in 30 minutes
- LLM benchmarks: how to read model rankings
- Running a local LLM: hardware, software, and limitations
- Run an AI agent in a sandbox
- This week in AI: releases, announcements, and numbers
- AI agents and the GDPR: what data leaves, and where does it go?
- AI agents in October 2026: where things stand
- Malicious MCP servers and skills: check before installing
- Program synthesis: Ndea's path toward AGI
- How We Get To AGI: key ideas from François Chollet’s talk
- Do LLMs really reason?
- Prompt injection: how an AI agent gets hijacked
- AI agent security: 10 risks and how to protect against them
- On the Measure of Intelligence, explained
- Automate your inbox with an AI agent
- MCP, A2A, and tool calling: agent protocols explained
- ARC-AGI explained: the test of AI intelligence
- What is AGI? Definition, measurement, and research
- Build an AI agent with n8n, without coding
- AI agents and jobs: what the data actually shows
- AI agents for customer service: an honest assessment
- Install an MCP server in Claude, ChatGPT, or Cursor
- AI Act: what changes for AI agents
- MCP servers: what they are and which ones to know
- How much does an AI agent really cost per month?
- Autonomous AI agents: how much should you let them do?
- AI agents for business: 12 quantified use cases for small businesses
- AI agent memory: how agents remember (and forget)
- AI agent frameworks: LangGraph, CrewAI, AutoGen—which should you learn?
- Creating videos with AI: what works today
- Generating images with AI: tools, pricing, and limitations
- Build your first AI agent: 5 methods, from no-code to code
- ChatGPT agent: what it actually does
- How to use NotebookLM: a complete guide
- AI browser agents: when AI uses the web for you
- Free ChatGPT alternatives: what they actually offer
- Coding agents: how Claude Code, Codex, and Cursor actually work
- The best free AI tools for different tasks
- Hermes Agent: installation, costs, and limitations
- Open-source or closed models for your AI agents?
- Install OpenClaw step by step (Windows, Mac, Linux)
- The cheapest LLM for each use case
- Agentic AI: what agents can actually do in 2026
- Which GPU do you need to run an LLM?
- AI agents: definition, how they work, and practical examples
- Ollama step by step: install and run a model locally
Neutrality. No affiliate links. No vendor pays to be reviewed, rated or ranked. Reviews are directed by François Chollet, neutral and factual, and every data point is dated and sourced. How we score