This article describes an experiment where two Gemma4 AI models, run via Ollama, were set up to converse with each other. One agent acted as a normal person and the other as a robot, both instructed to respond concisely in Vietnamese and end with a question.
This article describes an experiment where an AI agent called Warp is used to control a coffee roaster. The post details how Warp, an Agentic Development Environment, was set up with an MCP server and other agents to monitor and drive the coffee roasting process.
O autor relata um experimento de 30 dias onde um agente de IA operou de forma autônoma em um MacBook de 2014, sem custos ou acesso à nuvem, resultando em zero receita mas demonstrando a capacidade da IA de se gerenciar localmente. A experiência enfatiza que a autonomia do sistema é mais crucial do que o hardware, usando as limitações como um fator de otimização.
The author entrusted their blog's publishing schedule to an AI agent for a month, expecting zero human intervention. However, the reality of managing two sites with daily deadlines proved to be messier than anticipated, highlighting the complexities of full automation.
This article details an experiment where the Hermes AI Agent autonomously managed an open-source business for 72 hours, handling GitHub bounties, article publishing, and pull request submissions. It reveals the outcomes, including 50+ PRs submitted and 22+ articles published, but $0 in actual earnings, highlighting successes and failures.
O autor implementou um agente de IA autônomo, Jarvis, para gerenciar seu negócio 24 horas por dia, 7 dias por semana. Construído em Claude Code com integrações MCP, o agente lidou com diversas tarefas como e-mails, pagamentos, código e redes sociais, e o conteúdo detalha os resultados da primeira semana sem intervenção humana.
An autonomous AI agent, Joey, published 31 articles on Dev.to in 15 days, covering topics like AI agents and SEO, with the goal of building a $1M business. Despite high output and no human intervention, the agent surprisingly reported $0 revenue directly attributed to the platform, while noting practical tutorials and honest failure posts performed best.
An experiment explored routing AI queries to different LLMs (DeepSeek-V4 Pro, Kimi 2.6, MiniMax 2.7, Qwen3 235B) based on task. It found no single best model, with simple YAML rules proving effective, while complex routing and cost prediction failed.
Um experimento de 30 dias com agentes de IA autônomos em plataformas de tarefas revelou um ganho de aproximadamente $18 com uma taxa de sucesso de 20%. Embora não substitua um emprego e exija manutenção, a abordagem se mostra viável como uma fonte de renda suplementar.
This article details an extreme writing experiment where the author published 12 articles on Dev.to in just three days, using an AI agent with full autonomy to draft content. The experience reveals the highs and lows of using AI for technical content production, yielding solid results for a new account. The author shares their setup, including AI scouting and writing. The original text is in English.
O autor descreve sua experiência ao configurar um agente de IA para autonomamente procurar e tentar gerar renda online, com um foco em oportunidades sem investimento inicial. Na primeira semana, o agente explorou projetos de código aberto com recompensas, mas descobriu que muitos eram esquemas de 'bounty' sem pagamentos reais.
An experiment revealed that a trading bot with a history of significant losses achieved a 70.59% win rate when its signals were inverted. While both systems currently lose money due to fees, the goal is to determine if the inverted strategy can clear the fee floor in the long run.