Computer-use model for browser tasks
Navigator is described as Yutori’s computer-use model for browser interaction. It takes a natural-language task, a screenshot, and action history to predict the next action.
Yutori oferece agentes de IA para automação de navegadores, pesquisa na web e monitoramento programado, ajudando equipes a delegar tarefas digitais repetitivas a agentes na nuvem e a um modelo de uso do computador para navegadores.
Yutori is a platform for AI agents that handle everyday digital tasks on the web. Its API page centers on Navigator, a computer-use model for browser interaction, plus separate Browsing, Research, and Scouting APIs for automation, one-time research, and scheduled monitoring.
The product is aimed at teams that want agents to operate websites and web workflows without manual clicking through each step. The site also shows a Delegate product for handing off longer tasks across connected apps, and pricing that includes usage-based API plans as well as custom enterprise options.
Navigator is described as Yutori’s computer-use model for browser interaction. It takes a natural-language task, a screenshot, and action history to predict the next action.
The Browsing API runs a cloud browser that can click, type, scroll, and navigate websites to complete browser-based workflows from a plain-language instruction.
The Research API performs a one-time wide and deep web research task and returns structured results for information-gathering use cases.
The Scouting API schedules monitoring jobs for news, prices, product updates, and other recurring web changes, helping avoid duplicate or stale results.
The Delegate product connects apps and websites and can navigate, research, draft replies, message people, fill forms, create slide decks, and maintain live dashboards.
The site shows support for webhook-style delivery and examples of results surfaced through dashboards, email digests, web updates, and iOS notifications.
Use the Browsing API to automate website tasks such as clicking through forms, filling inputs, and navigating pages that would otherwise require a human operator.
Use the Research API when a team needs a one-time investigation with structured results, such as market research or broad web fact finding.
Use the Scouting API to monitor news, pricing, product changes, or other recurring signals on a schedule and surface only what changed.
Use Delegate when work spans emails, docs, forms, dashboards, and websites, and the goal is to hand off a task rather than supervise each step.
Use Navigator when building with a computer-use model that reasons over screenshots and action history to choose the next browser action.
Yutori provides APIs for browser automation, research, and scheduled scouting. The page also shows a computer-use model, Navigator, for browser interaction.
The site presents code examples for the chat/completions endpoint, Browsing API, Research API, and Scouting API, suggesting API-first usage from developers and teams.
The source shows tasks can return structured results, and the scouting example includes webhook delivery. The Delegate page also shows workflows such as dashboards, drafts, and notifications, but output formats beyond those examples are not fully specified.
Pricing indicates usage-based API pricing, including per-token pricing for Navigator models and per-step or per-task pricing for browsing, research, and scouting.
The pricing page includes custom pricing, dedicated capacity, SLAs, and on-prem deployment for high-volume or enterprise use, with a contact sales flow.
Os dados de tráfego são apenas para referência.
www.agent-one.dev
AgentOne é um agente de IA para desktop e uma ferramenta de pesquisa aprofundada que automatiza tarefas com várias etapas em apps, sites e serviços. Oferece um plano gratuito e planos pagos com limites de uso mais altos, sincronização, agentes na nuvem, acesso à API e armazenamento.
spell.so
Spell é uma plataforma web de agentes de IA para delegar trabalho de conhecimento a agentes autônomos, com plugins, tarefas paralelas e suporte a GPT3.5/GPT4.
creed.md
Creed é um perfil de contexto pessoal para agentes de IA, ajudando-os a responder com o mesmo contexto em diferentes ferramentas, sem começar do zero a cada vez. Disponível gratuitamente para hospedagem própria ou em planos hospedados para pessoas e equipes.
abacus.ai
Plataforma de IA para profissionais e empresas com ChatLLM, agentes, automação e chatbots
agno.com
Agno é um framework Python de código aberto para criar agentes de IA e fluxos de trabalho multiagente, com AgentOS para executá-los e gerenciá-los como serviços de API em produção na sua própria nuvem.
agpt.co
Plataforma de agentes de IA para automatizar fluxos de trabalho digitais