https://julien-riel.com/ https://julien-riel.com/97-choses/ https://julien-riel.com/97-choses/00-preface/ https://julien-riel.com/97-choses/01-agents-are-not-magic-they-are-probability/ https://julien-riel.com/97-choses/02-the-prompt-is-the-architecture/ https://julien-riel.com/97-choses/03-your-agent-is-only-as-good-as-its-context/ https://julien-riel.com/97-choses/04-stop-anthropomorphizing-start-debugging/ https://julien-riel.com/97-choses/05-trust-the-output-not-the-reasoning/ https://julien-riel.com/97-choses/06-agents-fail-gracefully-or-they-dont-there-is-no-middle/ https://julien-riel.com/97-choses/07-the-human-in-the-loop-is-a-feature-not-a-weakness/ https://julien-riel.com/97-choses/08-know-when-to-use-an-agent-and-when-to-use-a-function/ https://julien-riel.com/97-choses/09-determinism-is-a-choice-you-have-to-make-on-purpose/ https://julien-riel.com/97-choses/10-your-agent-has-no-memory-unless-you-give-it-one/ https://julien-riel.com/97-choses/11-give-your-agent-a-role-not-just-a-task/ https://julien-riel.com/97-choses/12-ambiguity-is-your-problem-not-the-agents/ https://julien-riel.com/97-choses/13-the-specification-is-the-skill/ https://julien-riel.com/97-choses/14-review-agent-output-like-you-review-a-juniors-pull-request/ https://julien-riel.com/97-choses/15-conversation-is-a-development-environment/ https://julien-riel.com/97-choses/16-prompts-drift-version-them-like-code/ https://julien-riel.com/97-choses/17-examples-outperform-instructions/ https://julien-riel.com/97-choses/18-negative-space-matters-tell-your-agent-what-not-to-do/ https://julien-riel.com/97-choses/19-system-prompts-are-contracts/ https://julien-riel.com/97-choses/20-the-best-prompt-is-the-one-you-dont-have-to-change/ https://julien-riel.com/97-choses/21-few-shot-is-not-fine-tuning/ https://julien-riel.com/97-choses/22-chain-of-thought-is-a-debugging-tool-not-just-a-performance-trick/ https://julien-riel.com/97-choses/23-prompting-is-thinking-out-loud-so-think-carefully/ https://julien-riel.com/97-choses/24-the-agent-that-sounds-confident-is-not-necessarily-correct/ https://julien-riel.com/97-choses/25-learn-to-recognize-hallucination-patterns-in-your-domain/ https://julien-riel.com/97-choses/26-design-for-observability-before-you-design-for-capability/ https://julien-riel.com/97-choses/27-evals-are-your-test-suite/ https://julien-riel.com/97-choses/28-the-tool-is-the-interface/ https://julien-riel.com/97-choses/29-idempotency-matters-more-in-agentic-systems-than-anywhere-else/ https://julien-riel.com/97-choses/30-dont-let-your-agent-touch-production-until-its-bored-you-in-staging/ https://julien-riel.com/97-choses/31-small-agents-beat-big-agents/ https://julien-riel.com/97-choses/32-orchestration-is-just-plumbing-treat-it-that-way/ https://julien-riel.com/97-choses/33-state-is-the-hardest-problem-in-agentic-programming/ https://julien-riel.com/97-choses/34-the-retry-loop-is-where-systems-go-to-die/ https://julien-riel.com/97-choses/35-your-agent-needs-a-kill-switch/ https://julien-riel.com/97-choses/36-log-everything-your-agent-was-thinking-not-just-what-it-did/ https://julien-riel.com/97-choses/37-timeouts-are-not-optional/ https://julien-riel.com/97-choses/38-cost-is-an-architectural-constraint/ https://julien-riel.com/97-choses/39-context-windows-are-budgets-spend-them-wisely/ https://julien-riel.com/97-choses/40-the-best-agents-have-a-narrow-personality/ https://julien-riel.com/97-choses/41-users-will-break-your-agent-in-ways-you-cannot-predict/ https://julien-riel.com/97-choses/42-latency-is-a-ux-problem-not-just-an-infrastructure-problem/ https://julien-riel.com/97-choses/43-never-let-an-agent-send-an-email-it-cannot-unsend/ https://julien-riel.com/97-choses/44-scope-creep-kills-agents-define-the-mission-narrowly/ https://julien-riel.com/97-choses/45-multi-agent-systems-multiply-capability-and-multiply-failure-modes/ https://julien-riel.com/97-choses/46-the-agent-that-does-everything-does-nothing-well/ https://julien-riel.com/97-choses/47-security-starts-with-what-you-put-in-the-context-window/ https://julien-riel.com/97-choses/48-prompt-injection-is-the-new-sql-injection/ https://julien-riel.com/97-choses/49-your-agent-will-agree-with-you-thats-the-problem/ https://julien-riel.com/97-choses/50-switching-models-is-switching-collaborators/ https://julien-riel.com/97-choses/51-know-what-your-agent-cannot-know/ https://julien-riel.com/97-choses/52-working-with-agents-gets-better-when-you-get-better-at-writing/ https://julien-riel.com/97-choses/53-you-are-responsible-for-everything-the-agent-does/ https://julien-riel.com/97-choses/54-you-are-the-senior-developer-the-agent-is-the-junior/ https://julien-riel.com/97-choses/55-agentic-programming-rewards-the-lazy-thinker/ https://julien-riel.com/97-choses/56-the-goal-is-outcomes-not-outputs/ https://julien-riel.com/97-choses/57-automate-the-boring-parts-stay-close-to-the-interesting-parts/ https://julien-riel.com/97-choses/58-iteration-speed-is-your-competitive-advantage/ https://julien-riel.com/97-choses/59-the-field-is-moving-your-mental-models-must-too/ https://julien-riel.com/97-choses/60-learn-to-read-failure-like-a-detective-not-a-judge/ https://julien-riel.com/97-choses/61-agentic-programming-is-a-discipline-not-a-shortcut/ https://julien-riel.com/97-choses/62-the-hardest-skill-is-knowing-when-to-take-back-the-wheel/ https://julien-riel.com/97-choses/63-discomfort-with-uncertainty-is-a-liability-in-this-field/ https://julien-riel.com/97-choses/64-expertise-still-matters-it-just-shows-up-differently-now/ https://julien-riel.com/97-choses/65-the-best-practitioners-are-editors-not-just-authors/ https://julien-riel.com/97-choses/66-patience-with-ambiguity-is-a-technical-skill/ https://julien-riel.com/97-choses/67-stay-curious-about-failure/ https://julien-riel.com/97-choses/68-agentic-systems-expose-gaps-in-your-own-thinking/ https://julien-riel.com/97-choses/69-the-field-rewards-generalists-who-go-deep-on-one-thing/ https://julien-riel.com/97-choses/70-dont-mistake-fluency-for-understanding/ https://julien-riel.com/97-choses/71-build-for-the-agent-you-have-not-the-agent-you-wish-you-had/ https://julien-riel.com/97-choses/72-the-first-version-should-be-embarrassingly-simple/ https://julien-riel.com/97-choses/73-your-ide-is-now-a-conversation/ https://julien-riel.com/97-choses/74-give-the-assistant-your-constraints-not-just-your-requirements/ https://julien-riel.com/97-choses/75-read-every-line-it-writes/ https://julien-riel.com/97-choses/76-the-assistant-doesnt-know-your-codebase-unless-you-show-it/ https://julien-riel.com/97-choses/77-use-it-to-understand-not-just-to-produce/ https://julien-riel.com/97-choses/78-commit-often-so-you-have-somewhere-to-return-to/ https://julien-riel.com/97-choses/79-the-best-use-of-an-ai-assistant-is-the-task-you-were-about-to-skip/ https://julien-riel.com/97-choses/80-dont-let-the-assistant-drive-the-architecture/ https://julien-riel.com/97-choses/81-context-is-a-skill-you-can-improve/ https://julien-riel.com/97-choses/82-an-ai-pair-programmer-has-no-ego-use-that/ https://julien-riel.com/97-choses/83-start-your-prompt-with-the-outcome-not-the-method/ https://julien-riel.com/97-choses/84-show-the-assistant-what-good-looks-like-in-your-codebase/ https://julien-riel.com/97-choses/85-when-the-output-is-wrong-fix-the-prompt-before-you-fix-the-code/ https://julien-riel.com/97-choses/86-break-large-tasks-into-prompts-not-just-steps/ https://julien-riel.com/97-choses/87-tell-the-assistant-what-to-preserve-not-just-what-to-change/ https://julien-riel.com/97-choses/88-use-the-assistant-to-pressure-test-your-own-ideas/ https://julien-riel.com/97-choses/89-large-projects-need-a-document-the-assistant-can-always-read/ https://julien-riel.com/97-choses/90-write-the-spec-before-you-write-the-prompt/ https://julien-riel.com/97-choses/91-let-the-assistant-write-the-plan-then-edit-it/ https://julien-riel.com/97-choses/92-use-markdown-not-prose-for-specifications/ https://julien-riel.com/97-choses/93-treat-your-claude-md-like-a-hiring-document/ https://julien-riel.com/97-choses/94-break-the-project-into-phases-the-assistant-can-complete/ https://julien-riel.com/97-choses/95-keep-a-decision-log-the-assistant-can-reference/ https://julien-riel.com/97-choses/96-let-the-tests-define-the-contract-then-let-the-assistant-fill-it/ https://julien-riel.com/97-choses/97-the-bigger-the-project-the-more-you-need-to-stay-in-charge/ https://julien-riel.com/articles/ https://julien-riel.com/articles/advanced-prompt-engineering/ https://julien-riel.com/articles/context-window/ https://julien-riel.com/articles/evals/ https://julien-riel.com/articles/fine-tuning-vs-rag/ https://julien-riel.com/articles/multi-agent-architectures/ https://julien-riel.com/articles/rag/ https://julien-riel.com/articles/understanding-llms/ https://julien-riel.com/case-studies/ https://julien-riel.com/case-studies/cursor/ https://julien-riel.com/case-studies/duolingo/ https://julien-riel.com/case-studies/klarna/ https://julien-riel.com/case-studies/whoop-coach/ https://julien-riel.com/confidentialite/ https://julien-riel.com/cross-reference/ https://julien-riel.com/en/ https://julien-riel.com/en/97-things/ https://julien-riel.com/en/97-things/00-preface/ https://julien-riel.com/en/97-things/01-agents-are-not-magic-they-are-probability/ https://julien-riel.com/en/97-things/02-the-prompt-is-the-architecture/ https://julien-riel.com/en/97-things/03-your-agent-is-only-as-good-as-its-context/ https://julien-riel.com/en/97-things/04-stop-anthropomorphizing-start-debugging/ https://julien-riel.com/en/97-things/05-trust-the-output-not-the-reasoning/ https://julien-riel.com/en/97-things/06-agents-fail-gracefully-or-they-dont-there-is-no-middle/ https://julien-riel.com/en/97-things/07-the-human-in-the-loop-is-a-feature-not-a-weakness/ https://julien-riel.com/en/97-things/08-know-when-to-use-an-agent-and-when-to-use-a-function/ https://julien-riel.com/en/97-things/09-determinism-is-a-choice-you-have-to-make-on-purpose/ https://julien-riel.com/en/97-things/10-your-agent-has-no-memory-unless-you-give-it-one/ https://julien-riel.com/en/97-things/11-give-your-agent-a-role-not-just-a-task/ https://julien-riel.com/en/97-things/12-ambiguity-is-your-problem-not-the-agents/ https://julien-riel.com/en/97-things/13-the-specification-is-the-skill/ https://julien-riel.com/en/97-things/14-review-agent-output-like-you-review-a-juniors-pull-request/ https://julien-riel.com/en/97-things/15-conversation-is-a-development-environment/ https://julien-riel.com/en/97-things/16-prompts-drift-version-them-like-code/ https://julien-riel.com/en/97-things/17-examples-outperform-instructions/ https://julien-riel.com/en/97-things/18-negative-space-matters-tell-your-agent-what-not-to-do/ https://julien-riel.com/en/97-things/19-system-prompts-are-contracts/ https://julien-riel.com/en/97-things/20-the-best-prompt-is-the-one-you-dont-have-to-change/ https://julien-riel.com/en/97-things/21-few-shot-is-not-fine-tuning/ https://julien-riel.com/en/97-things/22-chain-of-thought-is-a-debugging-tool-not-just-a-performance-trick/ https://julien-riel.com/en/97-things/23-prompting-is-thinking-out-loud-so-think-carefully/ https://julien-riel.com/en/97-things/24-the-agent-that-sounds-confident-is-not-necessarily-correct/ https://julien-riel.com/en/97-things/25-learn-to-recognize-hallucination-patterns-in-your-domain/ https://julien-riel.com/en/97-things/26-design-for-observability-before-you-design-for-capability/ https://julien-riel.com/en/97-things/27-evals-are-your-test-suite/ https://julien-riel.com/en/97-things/28-the-tool-is-the-interface/ https://julien-riel.com/en/97-things/29-idempotency-matters-more-in-agentic-systems-than-anywhere-else/ https://julien-riel.com/en/97-things/30-dont-let-your-agent-touch-production-until-its-bored-you-in-staging/ https://julien-riel.com/en/97-things/31-small-agents-beat-big-agents/ https://julien-riel.com/en/97-things/32-orchestration-is-just-plumbing-treat-it-that-way/ https://julien-riel.com/en/97-things/33-state-is-the-hardest-problem-in-agentic-programming/ https://julien-riel.com/en/97-things/34-the-retry-loop-is-where-systems-go-to-die/ https://julien-riel.com/en/97-things/35-your-agent-needs-a-kill-switch/ https://julien-riel.com/en/97-things/36-log-everything-your-agent-was-thinking-not-just-what-it-did/ https://julien-riel.com/en/97-things/37-timeouts-are-not-optional/ https://julien-riel.com/en/97-things/38-cost-is-an-architectural-constraint/ https://julien-riel.com/en/97-things/39-context-windows-are-budgets-spend-them-wisely/ https://julien-riel.com/en/97-things/40-the-best-agents-have-a-narrow-personality/ https://julien-riel.com/en/97-things/41-users-will-break-your-agent-in-ways-you-cannot-predict/ https://julien-riel.com/en/97-things/42-latency-is-a-ux-problem-not-just-an-infrastructure-problem/ https://julien-riel.com/en/97-things/43-never-let-an-agent-send-an-email-it-cannot-unsend/ https://julien-riel.com/en/97-things/44-scope-creep-kills-agents-define-the-mission-narrowly/ https://julien-riel.com/en/97-things/45-multi-agent-systems-multiply-capability-and-multiply-failure-modes/ https://julien-riel.com/en/97-things/46-the-agent-that-does-everything-does-nothing-well/ https://julien-riel.com/en/97-things/47-security-starts-with-what-you-put-in-the-context-window/ https://julien-riel.com/en/97-things/48-prompt-injection-is-the-new-sql-injection/ https://julien-riel.com/en/97-things/49-your-agent-will-agree-with-you-thats-the-problem/ https://julien-riel.com/en/97-things/50-switching-models-is-switching-collaborators/ https://julien-riel.com/en/97-things/51-know-what-your-agent-cannot-know/ https://julien-riel.com/en/97-things/52-working-with-agents-gets-better-when-you-get-better-at-writing/ https://julien-riel.com/en/97-things/53-you-are-responsible-for-everything-the-agent-does/ https://julien-riel.com/en/97-things/54-you-are-the-senior-developer-the-agent-is-the-junior/ https://julien-riel.com/en/97-things/55-agentic-programming-rewards-the-lazy-thinker/ https://julien-riel.com/en/97-things/56-the-goal-is-outcomes-not-outputs/ https://julien-riel.com/en/97-things/57-automate-the-boring-parts-stay-close-to-the-interesting-parts/ https://julien-riel.com/en/97-things/58-iteration-speed-is-your-competitive-advantage/ https://julien-riel.com/en/97-things/59-the-field-is-moving-your-mental-models-must-too/ https://julien-riel.com/en/97-things/60-learn-to-read-failure-like-a-detective-not-a-judge/ https://julien-riel.com/en/97-things/61-agentic-programming-is-a-discipline-not-a-shortcut/ https://julien-riel.com/en/97-things/62-the-hardest-skill-is-knowing-when-to-take-back-the-wheel/ https://julien-riel.com/en/97-things/63-discomfort-with-uncertainty-is-a-liability-in-this-field/ https://julien-riel.com/en/97-things/64-expertise-still-matters-it-just-shows-up-differently-now/ https://julien-riel.com/en/97-things/65-the-best-practitioners-are-editors-not-just-authors/ https://julien-riel.com/en/97-things/66-patience-with-ambiguity-is-a-technical-skill/ https://julien-riel.com/en/97-things/67-stay-curious-about-failure/ https://julien-riel.com/en/97-things/68-agentic-systems-expose-gaps-in-your-own-thinking/ https://julien-riel.com/en/97-things/69-the-field-rewards-generalists-who-go-deep-on-one-thing/ https://julien-riel.com/en/97-things/70-dont-mistake-fluency-for-understanding/ https://julien-riel.com/en/97-things/71-build-for-the-agent-you-have-not-the-agent-you-wish-you-had/ https://julien-riel.com/en/97-things/72-the-first-version-should-be-embarrassingly-simple/ https://julien-riel.com/en/97-things/73-your-ide-is-now-a-conversation/ https://julien-riel.com/en/97-things/74-give-the-assistant-your-constraints-not-just-your-requirements/ https://julien-riel.com/en/97-things/75-read-every-line-it-writes/ https://julien-riel.com/en/97-things/76-the-assistant-doesnt-know-your-codebase-unless-you-show-it/ https://julien-riel.com/en/97-things/77-use-it-to-understand-not-just-to-produce/ https://julien-riel.com/en/97-things/78-commit-often-so-you-have-somewhere-to-return-to/ https://julien-riel.com/en/97-things/79-the-best-use-of-an-ai-assistant-is-the-task-you-were-about-to-skip/ https://julien-riel.com/en/97-things/80-dont-let-the-assistant-drive-the-architecture/ https://julien-riel.com/en/97-things/81-context-is-a-skill-you-can-improve/ https://julien-riel.com/en/97-things/82-an-ai-pair-programmer-has-no-ego-use-that/ https://julien-riel.com/en/97-things/83-start-your-prompt-with-the-outcome-not-the-method/ https://julien-riel.com/en/97-things/84-show-the-assistant-what-good-looks-like-in-your-codebase/ https://julien-riel.com/en/97-things/85-when-the-output-is-wrong-fix-the-prompt-before-you-fix-the-code/ https://julien-riel.com/en/97-things/86-break-large-tasks-into-prompts-not-just-steps/ https://julien-riel.com/en/97-things/87-tell-the-assistant-what-to-preserve-not-just-what-to-change/ https://julien-riel.com/en/97-things/88-use-the-assistant-to-pressure-test-your-own-ideas/ https://julien-riel.com/en/97-things/89-large-projects-need-a-document-the-assistant-can-always-read/ https://julien-riel.com/en/97-things/90-write-the-spec-before-you-write-the-prompt/ https://julien-riel.com/en/97-things/91-let-the-assistant-write-the-plan-then-edit-it/ https://julien-riel.com/en/97-things/92-use-markdown-not-prose-for-specifications/ https://julien-riel.com/en/97-things/93-treat-your-claude-md-like-a-hiring-document/ https://julien-riel.com/en/97-things/94-break-the-project-into-phases-the-assistant-can-complete/ https://julien-riel.com/en/97-things/95-keep-a-decision-log-the-assistant-can-reference/ https://julien-riel.com/en/97-things/96-let-the-tests-define-the-contract-then-let-the-assistant-fill-it/ https://julien-riel.com/en/97-things/97-the-bigger-the-project-the-more-you-need-to-stay-in-charge/ https://julien-riel.com/en/articles/ https://julien-riel.com/en/articles/advanced-prompt-engineering/ https://julien-riel.com/en/articles/context-window/ https://julien-riel.com/en/articles/evals/ https://julien-riel.com/en/articles/fine-tuning-vs-rag/ https://julien-riel.com/en/articles/multi-agent-architectures/ https://julien-riel.com/en/articles/rag/ https://julien-riel.com/en/articles/understanding-llms/ https://julien-riel.com/en/case-studies/ https://julien-riel.com/en/case-studies/cursor/ https://julien-riel.com/en/case-studies/duolingo/ https://julien-riel.com/en/case-studies/klarna/ https://julien-riel.com/en/case-studies/whoop-coach/ https://julien-riel.com/en/glossary/ https://julien-riel.com/en/privacy/ https://julien-riel.com/en/projets/ https://julien-riel.com/en/reading-guide/ https://julien-riel.com/glossaire/ https://julien-riel.com/projets/ https://julien-riel.com/projets/planificateur-terrasse/ https://julien-riel.com/reading-guide/ https://julien-riel.com/sites/ https://julien-riel.com/tags/ https://julien-riel.com/tags/agents-in-the-real-world/ https://julien-riel.com/tags/agents/ https://julien-riel.com/tags/ai-integration/ https://julien-riel.com/tags/architecture/ https://julien-riel.com/tags/book/ https://julien-riel.com/tags/building-agentic-systems/ https://julien-riel.com/tags/content-pipeline/ https://julien-riel.com/tags/context-engineering/ https://julien-riel.com/tags/customer-service/ https://julien-riel.com/tags/developer-as-user/ https://julien-riel.com/tags/developer-tools/ https://julien-riel.com/tags/evals/ https://julien-riel.com/tags/fine-tuning/ https://julien-riel.com/tags/fondamentaux/ https://julien-riel.com/tags/fundamentals/ https://julien-riel.com/tags/human-in-the-loop/ https://julien-riel.com/tags/mindset/ https://julien-riel.com/tags/multi-agent/ https://julien-riel.com/tags/primer/ https://julien-riel.com/tags/product-strategy/ https://julien-riel.com/tags/prompting-as-engineering/ https://julien-riel.com/tags/prompting/ https://julien-riel.com/tags/rag/ https://julien-riel.com/tags/testing/ https://julien-riel.com/tags/vulgarisation/ https://julien-riel.com/tags/working-with-agents/