Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.twin.so/llms.txt

Use this file to discover all available pages before exploring further.

​
General Questions

Twin is an AI-native automation platform that helps teams and individuals build, deploy, and manage workflows faster. Whether you’re a no-code creator or a developer, Twin turns plain English instructions into fully functional Rust-based (code) automations executed by AI agents.
Twin turns your natural language instructions into an executable AI agent built with code behind the scenes. All your automation steps and integrations are dynamically coded by a coding agent.
Absolutely not. Twin is designed for non-technical users. If your team has technical needs (advanced integrations, custom endpoints, deeper control), Twin can still fit into that environment — but you don’t need coding knowledge to get value.
Most users can get a first agent working quickly (often in minutes) for a simple workflow like “go to a URL → extract content → send it somewhere.” More complex processes may take a bit of iteration on the chat, especially if login steps, edge cases, or brittle websites are involved.Watch our tutorials:
Twin can work with ANY App. You can automate anything — from simple data tasks to complex, AI-powered workflows or AI Agents that run your business in the background.Typical use cases include:
  • Competitive monitoring (careers pages, pricing pages, product updates)
  • Lead research and enrichment from websites
  • Social media monitoring
  • Invoice retrieval from vendor portals
  • Workflows on software that doesn’t expose an API
Twin is especially useful when traditional automation breaks down because the job happens inside a web UI.
Traditional no-code tools rely on rigid step-by-step nodes, limited actions, and pre-built integrations library. Twin generates real Rust automation code, empowering Twin to handle complex logic and deep integrations, all without writing a single line of code. You can describe what you want in plain English, and Twin builds it for you.
Yes! Twin’s Interfaces feature lets you build full websites and apps — with sign-ups, databases, and your Twin agents powering the backend — directly inside Twin. You can also deploy agents to your own frontend via the REST API.
Yes. Twin offers a 14-day free trial with 1,000 credits to start and +200 credits per day during the trial. See Pricing for the current details.
The Orchestrator is the left-side chat in your workspace. It has full context on every agent, run, and schedule — and it’s how you create, modify, and delete agents using natural language. It will also proactively suggest splitting complex workflows into focused agents or creating new ones when it sees an opportunity.
Build mode is when you’re creating or updating an agent - it uses high-reasoning AI and is the most credit-intensive. Run mode is when a deployed agent executes - it reuses the agent’s tools and memory, so repeat runs are typically 3-10x cheaper. Build once, then run on schedule for maximum efficiency.
Twin never interrupts a build that is already in progress. During the trial, you keep receiving daily credits until the 14 days end; on paid plans, you can top up or wait for your next billing cycle. See Pricing for the current plan details.
No. When an agent needs browser automation, it runs in an isolated cloud computer dedicated to your agent — completely separate from your machine. Your local browser (Chrome, Safari, Arc) is never accessed. Think of it as a virtual computer we dedicate to you. See Web Agent for more.
Interfaces let you build full websites and apps — with sign-ups, databases, and your agents as the backend — directly inside Twin. Describe what you want in the chat, and Twin generates a complete, deployable application. You can publish it as Private, Restricted, or Public.
Yes. With Interfaces, you can build customer-facing apps with authentication and databases, powered by your Twin agents. Because you control the frontend, backend logic, and access — you can build products you charge for.
Common causes:
  • First builds are expensive - the agent is discovering APIs, handling auth, and shaping tools for the first time. Later runs reuse what it learned and are often 3-10x cheaper.
  • Retry loops — if a step keeps failing (e.g., sandbox restrictions or permission errors), the agent may retry repeatedly, burning credits without progress.
  • Context accumulation — long-running agents carry the full conversation history, so each step gets more expensive as the session grows.
  • Expensive tool choices — the agent may default to browser automation when an API would be cheaper and faster.
What to do: Simplify your goal, break complex workflows into multiple focused agents, prefer API-first instructions, and monitor early runs closely. See Tips and Tricks for detailed best practices.
This can happen when:
  • Conflicting constraints — goals with many conditions can create competing priorities. Simplify and be explicit about what matters most.
  • Technical fallbacks — when a preferred method fails, the agent may fall back to a tool you explicitly prohibited. Restate the prohibition clearly and provide an alternative.
  • Step limits — agents don’t always track internal counters reliably. Instead of “do at most 5 searches,” try “stop after finding the first 3 results.”
What to do: Use clear, specific language. Test incrementally with simple tasks first. If the agent consistently violates explicit instructions, contact live in-app support with the run ID — the team can investigate and adjust credits if needed.
Create an account, open the chat, and describe the workflow you want to automate. Start with a narrow task you can verify quickly (one website, one output destination), then expand from there.

Get Started Now

Create your free account

​
Features and Capabilities

Twin can integrate with ANY app. We don’t rely on a pre-built integrations library. Twin generates real Rust automation code to help you automate on any app.For the most common business tools (Gmail, Slack, Google Sheets, etc.) we have integrated Google OAuth to offer a seamless experience.The key difference is that Twin can operate directly inside web apps when APIs aren’t available, using our Web Browser Agent to control the interface.
Yes. You can ask the agent to connect using Lovable API or any other external service.
Yes. You can run agents when Slack messages, email arrivals, webhook events, or form submissions occur. You can also run agents on a time schedule or trigger them from your own backend via the REST API. See Triggers for all options.
No. Twin runs the agent executions in managed environments so you don’t have to provision infrastructure for typical deployments.
Yes. Inside a Workspace, the Orchestrator can create and coordinate multiple agents that share context. It will even proactively suggest splitting a complex workflow into focused agents. One agent’s output can feed into another automatically.
Each workspace supports one connected account per service (e.g., one X/Twitter account, one Gmail account). All agents in that workspace share the same connection. If you need to manage multiple accounts for the same service, create a separate workspace for each. Multi-account management — including managing accounts on behalf of your clients — is coming soon. See Workspaces for details.
Twin is built to handle multi-step workflows. For higher volume or longer-running jobs, you typically design the process with batching, retries, and monitoring in mind. Twin can surface run results and failures so you can iterate.

​
Running Workflows

Common options:
  • Run it manually from the Twin interface
  • Run it on a schedule
  • Trigger it from another system (webhook/event)
  • Trigger it programmatically via the REST API
Yes. Deleted agents can be restored manually by the Twin team. Reach out via live in-app support with the name of the agent and the team will recover it for you.
Depending on your workspace settings, you can share agents internally with teammates, and in some cases expose them as templates or reusable automations.
You can specify it directly in your initial prompt, or configure it manually after building. Twin supports cron-based schedules, event-based triggers, and programmatic triggers via the REST API. See Triggers for details.
Yes — you can share workflows by clicking Share agent in the right panel. Anyone with the link can view the step-by-step run and clone the workflow in one click.

​
Support

Twin has live in-app support with real human agents from the team — average response time is under 20 minutes. Open the support chat directly inside Twin and you’ll be talking to someone on the team.You can also browse our documentation or watch video tutorials on our YouTube channel.
The fastest way is through the live in-app support inside Twin — real humans from the team reply in under 20 minutes. For anything else, email hugo@twin.so (CEO).
Open Twin and use the live in-app support chat. Human agents from the Twin team are available and typically respond in under 20 minutes. You can also email hugo@twin.so.
Twin provides full run visibility so you can see exactly what happened. Use the live in-app support inside Twin to share run details — the team will help you debug quickly, usually responding in under 20 minutes.

​
About Twin

Yes — many teams use Twin to deliver automation outcomes for customers or internal stakeholders. If you’re interested in distributing Twin to your clients or becoming a partner, reach out to alessandro@twin.so.
We’re open to partnerships, reseller agreements, and distribution deals. Contact alessandro@twin.so to discuss how we can work together.
Yes — Twin’s Interfaces feature lets you build and publish full websites and apps directly inside Twin, with your agents as the backend. No external tools needed. You can also choose to make your app Private, Restricted, or Public.
Yes. You can create agents, trigger runs, and manage schedules programmatically. See the full REST API documentation for endpoints and examples.
Go to your Twin account and navigate to API key management, or create one via the REST API itself.
Yes. Security is built into Twin by design. API tokens and keys are encrypted at rest and only decrypted in memory when needed for a request. Agent runs execute in isolated sandboxes, and credential access is tightly controlled, authenticated, and audited across internal services.
If you discover a potential security vulnerability, please report it responsibly by emailing joao@twin.so with the following details:
  • Impact assessment — what are the consequences (e.g., unauthorized access, data exposure)
  • Timeline — when the vulnerability was first observed
  • Technical details — logs, affected Agent IDs, or any specifics that help our team investigate
Our security and engineering teams will review your report and follow up regarding next steps and any potential rewards.
100% Rust :)
Yes. Contact us via live in-app support and provide your company information (name, address, VAT number, and any other details you need on the invoice). The team will issue it for you.
Yes! You can earn rewards for referring new customers to Twin. Sign up for the affiliate program here:

Join the Affiliate Program

Sign up and start earning
Check out the Changelog where we document our research, engineering and product work.Follow us on LinkedIn, X, and YouTube if you don’t want to miss any updates.

​
Still Have Questions?

Live In-App Support

Real humans from the team — response in under 20 minutes

Book a Call

Talk to the founding team