Twin

What is Twin?

Twin is a no-code platform that turns a written job description into an AI agent, which then carries out recurring business tasks across your apps without being prompted each time.

Setting one up feels like a conversation. You explain the outcome you want, Twin asks clarifying questions, and it assembles an agent with the right app connections, steps and schedule. Each run produces a report, and anything the agent is unsure about is flagged for a person to review. Where an app has no API, the agent opens a virtual browser, logs in and works through the pages itself.

Twin Labs, based in Paris, was founded by Hugo Mercier and João Justi. The agent platform launched publicly in January 2026 alongside a reported $10 million seed round, and the homepage now claims 100,000 users and three million agents built. Its site showed current plans and a recent changelog during our October 2026 review.

Key Features

Twin combines a conversational agent builder with a browser agent, which lets one automation span both connected apps and websites that offer no integration.

• Conversational builder: You describe a task in plain English and Twin drafts the agent, asking follow-up questions where the brief is vague. More than 250 ready-made templates cover jobs such as lead follow-up, review replies and weekly reporting.
• Browser agent: On the Pro plan, agents can log in to websites and click through forms when no API exists. Since late 2025 the thread shows every action taken on a page along with the agent’s reasoning.
• App connections: Twin claims links to more than 5,000 apps, with Gmail, Slack, Notion, Google Sheets, Google Docs and HubSpot shown first. Older documentation quotes 2,700, so check that the tools you need are listed.
• Schedules and triggers: Agents run on a timetable written in ordinary words, such as every morning at 09:00, or fire from webhooks like a new email or form entry. Developers can also start runs through the REST API.
• Approvals and shared memory: Sensitive steps can pause for a draft-and-approve check before anything is sent. Agents on the same team share memory, so one agent can reuse context that another has already gathered.
• Customer dashboards: Agencies can run agents on behalf of clients and give each client a dashboard of results. Plus and Pro include up to five dashboards, while the Store plan adds per-customer tuning at scale.

Pros

Flat-rate pricing with unlimited agents is Twin’s strongest selling point, because teams can automate heavily without watching a credit balance fall.

• No coding needed: Sales staff, estate agents and office managers can build working agents by describing the job. The builder chooses apps and steps, so nobody has to design a flowchart.
• Unlimited runs: Every plan includes unlimited agents and scheduled runs with no credit meter or overage fees. High-volume tasks such as daily lead checks cost the same as occasional ones.
• Stated security credentials: Twin’s pricing page lists ISO 27001 certification and SOC 2 Type I and Type II reports. A public status page and a subprocessor list are also available for procurement checks.
• Trial without a card: The seven-day free trial needs no payment details. You can build and test a real agent on your own accounts before deciding whether to subscribe.
• Clear run records: Each run is logged with a report, and items the agent is unsure about are flagged for review. This makes it easier to spot mistakes before they reach customers.

Cons

Browser automation sits behind the $149 Pro plan, so the feature that most separates Twin from rivals costs five times the entry price.

• Steep plan jump: Plus at $29 covers app-based agents only, and the browser agent, prospecting and research need Pro at $149. Small teams should confirm which tier their main task requires before the trial ends.
• No free plan: Twin offers a seven-day trial rather than a permanent free tier. Older documentation mentions 1,000 free credits, but the current pricing page does not offer them.
• Agents still make errors: Autonomous agents can misread a page or pick the wrong record, particularly in the first few runs. Use approval steps and review outputs daily until you trust a workflow.
• Browser runs are slower: Working through web pages takes longer than an API call, and sites that change their layout can interrupt a run. Time-critical jobs are better routed through a native integration.
• Conflicting public figures: Different Twin pages quote 2,700 or 5,000 apps, and one guide prices Pro in euros. These leftovers make it harder to judge the product from search results alone.

Who is Using Twin?

Small businesses with repetitive web and inbox work form Twin’s core audience, especially in sales, property, recruiting and agency services.

• Sales teams: Reps use agents to find prospects on LinkedIn, enrich contacts and send follow-ups. Twin’s demo shows one agent replying to a new lead in 41 seconds.
• Property professionals: Estate agents and investors gather seller leads and pull property records. Some agents also research foreclosures from public court filings.
• Operations and finance staff: Back-office teams automate invoice collection, timesheets and record updates. A reported Qonto partnership used Twin to fetch supplier invoices from vendor portals.
• Marketing agencies: Agencies run agents for clients, handling review replies, reports and social posts. Client dashboards let each customer see results without logging in to Twin.

Pricing

Twin charges from $29 a month after a seven-day free trial that needs no card (official prices, October 2026).

A Twin guide page still quotes Pro from EUR 20 and the documentation mentions free credits, but neither appears on the current pricing page.

Disclaimer: Twin has changed its plan structure more than once, so recheck the pricing page on the day you subscribe.

What Makes Twin Unique?

A written brief replaces the flowchart in Twin, which is the opposite of how Zapier, Make and n8n ask you to build automations. The agent plans the steps itself and falls back to a virtual browser when an app has no connector, so one automation can mix a CRM, an inbox and a supplier portal.

Twin is also model-agnostic. Its homepage reports that routing routine steps to DeepSeek delivered about 96 per cent of Claude Opus quality at roughly 60 per cent of the cost, which helps explain how it can offer unlimited runs on a flat fee.

Compatibilities and Integrations

Twin runs in the browser and connects to thousands of apps, with Slack serving as a second place to manage agents.

• Platforms: Twin is a web app with keyboard navigation through CMD or CTRL+K. Agents can also be reached and configured from Slack channels.
• Apps: Named connections include Gmail, Google Sheets, Google Docs, Notion, HubSpot, Airtable, Shopify, Apollo and GoHighLevel. Twin claims more than 5,000 in total.
• Developers: A REST API and webhooks, documented at docs.twin.so, let you start agents from your own systems. This suits teams embedding Twin inside an existing product.
• Browser: A remote browser handles sites without integrations, and you can log in to accounts directly when the agent asks. Credentials are requested at run time rather than typed into prompts.

Twin Tutorials

Most people get a first agent running in under 15 minutes by starting from a narrow, repeatable task.

1. Start the seven-day trial at twin.so without entering a card.
2. Pick a template or describe one task, such as a weekly competitor summary.
3. Answer Twin’s follow-up questions and connect the apps it requests.
4. Review the proposed plan, add an approval step, then run a test.
5. Set a schedule in plain words or attach a webhook trigger.
6. Check run reports daily and turn on Slack notifications for that agent.

How We Rated It

Twin earns 4.0 out of 5, with broad integrations and a capable browser agent held back by the price of the Pro tier and the need to check early runs.

Summary

Twin suits small businesses, sales teams and agencies that want repetitive web and inbox work handled by agents rather than people. Its conversational builder, unlimited runs and browser agent make it one of the more practical no-code options we have tested for mixed app and website workflows. The main limitation is cost, because browser automation only arrives on the $149 Pro plan. Autonomous agents can also make mistakes, so approval steps matter for anything customer-facing. Use the free week to automate one narrow task end to end and compare the time saved against the plan you would need.