Learn
Short, citeable explanations of the concepts behind Twin: agents, browser automation, computer use, agentic workflows, last-mile AI, and the product definition of Twin itself.
Twin definition
A definition-first page with structured numbers for integrations, build time, trigger types, and pricing.
Read definitionAI agent
An AI agent is software that uses an AI model, instructions, tools, and memory to plan and complete tasks on behalf of a person or team.
Read definitionBrowser automation
Browser automation is software control of a web browser to visit sites, click buttons, fill forms, extract data, and complete tasks that normally require a human.
Read definitionComputer-use agent
A computer-use agent is an AI agent that operates software through a computer interface, using actions like clicking, typing, reading screens, opening files, and navigating apps.
Read definitionAgentic workflow
An agentic workflow is a business process where an AI agent plans steps, uses tools, evaluates results, and adapts its next action until the goal is complete.
Read definitionLast-mile AI
Last-mile AI is the execution layer that turns AI reasoning into finished business work across the tools, websites, files, and communication channels a team actually uses.
Read definitionGlossary
Definitions for the language users search before they know which product category they need.
Core concept
An AI agent is software that can interpret a goal, choose actions, use tools, and report an outcome. In business automation, agents usually connect to apps, APIs, websites, and documents so they can complete multi-step work rather than only answer questions.
Related: autonomous agent, workflow automation, tool use
An autonomous agent can run a planned workflow on its own, recover from routine errors, and ask for help only when human judgment or permission is needed. Autonomy is especially useful for recurring work such as daily reports, lead research, and CRM hygiene.
Related: scheduled agent, human-in-the-loop
Workflow automation coordinates repeated work across systems. Traditional tools require users to wire each trigger and action. AI-agent automation lets users describe the outcome while the agent plans tool calls, handles exceptions, and reports progress.
Related: AI agent, trigger, API connector
Capability
Browser automation lets an agent operate websites the way a human would: logging in, navigating pages, reading tables, filling forms, and downloading files. It is useful when the target system does not have an API or when the needed data is behind a login.
Related: web scraping, login-walled data
An API connector handles authentication, request formats, response parsing, and error behavior for an application. Connectors let agents update CRMs, send emails, create tasks, enrich records, and sync data without manual copy-paste.
Related: tool use, workflow automation
Run pattern
A scheduled agent wakes up at a set cadence, completes the same job with fresh context, and sends the result to the right place. Examples include a 7 AM market brief, a weekly competitive-intel scan, or a 15-minute support queue check.
Related: autonomous agent, trigger
A trigger is the condition that starts work. Common triggers include schedules, webhooks, inbound emails, new CRM records, form submissions, Slack messages, or a user typing a chat instruction.
Related: scheduled agent, webhook
Governance
Human-in-the-loop automation keeps people involved at decision points that need judgment, compliance review, or brand approval. An agent might draft outreach automatically but wait for a human before sending it.
Related: autonomous agent, approval workflow
AI infrastructure
Model routing sends each task to the model best suited for it. A fast model might classify simple records, while a stronger reasoning model handles ambiguous research, drafting, or troubleshooting.
Related: AI agent, tool use
Operations
A run log captures the actions, tool calls, outputs, errors, and decisions from an agent run. It helps teams understand what happened, debug failures, and trust that recurring work is being handled correctly.
Related: observability, workflow automation
Operating model
An AI workforce is a set of specialized agents that cover multiple functions, such as lead research, customer follow-up, content production, reporting, and back-office coordination. The point is not one magic assistant; it is many reliable agents doing narrow jobs.
Related: AI agent, autonomous agent
Pricing
Credits are a usage-based pricing unit. Instead of paying for every static step in a visual workflow, credit-based products typically charge when the agent performs real work such as model calls, tool calls, data extraction, or browser automation.
Related: workflow automation, AI agent
Get the latest product updates, tips, and insights delivered straight to your inbox.
No spam, unsubscribe anytime. We respect your privacy.