How much does Twin cost?

Credit based plans (Build Mode versus Run Mode credits) with a custom enterprise tier (custom credit volume, private support, commercial terms); specific plan rates not captured in retrieved materials.

Back to vendors

Twin

Q: Who is Twin for?

operations teams, SMBs and mid-market, fintech and platform partners, small teams building agent fleets

Also known as: Twin Labs, Twin A1, build.twin.so, Invoice Operator

Visit site

Browser / computer-use agentindependentVerified 2026-07-22

Autonomous AI employee: agents plan, choose tools, and execute across 50,000 apps and websites plus a proprietary browser agent for API-less portals, triggered by email, call, schedule, or webhook, with approval gates on irreversible actions. Proven at scale via Qonto's Invoice Operator for 500,000+ SMBs.

Twin builds autonomous AI agents that take a goal, plan the steps, choose tools, and execute across apps and websites until the workflow is complete, positioning itself as an AI employee rather than a copilot. Agents trigger themselves from an email, a call, a schedule, or a webhook; approval is required on anything irreversible and teams decide which actions run automatically versus which require review. The proprietary Browser agent handles the long tail where APIs do not exist, logging into legacy ERPs, scraping vendor portals, monitoring government sites, and pulling PDFs from customer dashboards, alongside thousands of prebuilt app connectors. The platform tracks cost and usage per agent, accepts live feedback to adjust runs, offers web search and deep research, and lets teams share agents internally or publish them to a marketplace with deployment and monetization. Twin Labs (founded 2024, France) proved the approach at production scale before the platform pivot: its Invoice Operator, built with Qonto and OpenAI, serves 500,000+ European SMBs retrieving invoices from portals, and Twin was one of 15 companies alpha testing OpenAI's Computer-Using Agent model. Credit based pricing across Build Mode and Run Mode with custom enterprise plans.

Vendor details

Canonical URL

https://twin.so

Funding status

Founded 2024 (Twin Labs, France); one of 15 companies chosen to alpha test OpenAI's Computer-Using Agent model; Invoice Operator with Qonto serves 500,000+ European SMBs

Company status

independent

Use cases & customers

Primary use cases

autonomous business workflow agentsbrowser automation of portals and legacy systems without APIsinvoice and document retrieval at scalelead research and CRM updatescompetitive monitoring and reporting

Target customers

operations teamsSMBs and mid-marketfintech and platform partnerssmall teams building agent fleets

Deployment options

cloud platform (build.twin.so)

Integrations

Thousands of prebuilt app connectors (vendor cites 5,000+ connectors and lists 50,000 apps and websites available) spanning CRMs, ERPs, marketing, communication, data warehouses, and e-commerce; the proprietary Browser agent logs into legacy ERPs, vendor portals, government sites, and gated dashboards where no API exists; vault system for credential management.

Sources & related URLs

Research sources

https://twin.so/ https://twin.so/features/autonomous-ai-agents/ https://twin.so/pricing https://tallyfy.com/products/pro/integrations/computer-ai-agents/vendors/twin/

Capability coverage

9.5 / 14 capabilities · 68%

Integrations & Tool CallingThousands of prebuilt app connectors (vendor cites 5,000+ and lists 50,000 apps and websites) plus the proprietary Browser agent for portals and legacy systems without APIs, Twin materials 2026-07-22	Full
Workflow OrchestrationAgents take a goal, split it into steps, call tools, browse websites, update systems, and continue until the workflow is complete, confirmed at production scale by the Qonto Invoice Operator, Twin materials and Tallyfy profile 2026-07-22	Full
Knowledge Grounding & RAGWeb search and deep research ship as plan features and agents ground on provided context like FAQs, without a documented retrieval architecture, Twin materials 2026-07-22	Partial
Human Oversight & GuardrailsApproval is required on anything irreversible, sensitive actions can include approval steps, and teams decide what runs automatically versus what requires review, Twin materials 2026-07-22	Full
Security, Identity & GovernanceA vault system stores website login credentials securely and the pricing page states audited and certified, without named attestations in retrieved materials, Twin and Tallyfy materials 2026-07-22	Partial
Observability & AuditabilityCost and usage tracking per agent and live feedback into running agents are documented, short of run level audit tooling, Twin materials 2026-07-22	Partial
Memory & State PersistenceAgents remember workflow context across runs per first party materials, without a documented memory architecture, Twin materials 2026-07-22	Partial
Deployment & Data ResidencyCloud platform only; no self hosting or data residency options documented, Twin materials 2026-07-22	Unable to verify
Prebuilt Agents, Templates & PacksTeams share agents internally or publish them to a marketplace, agents are reused across the team, and named use case templates ship on solution pages, Twin materials 2026-07-22	Full
Triggers & Channel CoverageAgents trigger themselves from an email, a call, a schedule, or webhook style triggers, in addition to manual runs, Twin materials 2026-07-22	Full
Model Flexibility & RoutingEvery latest model tested and ready per first party materials, indicating multi model access without documented customer routing control, Twin materials 2026-07-22	Partial
APIs, SDKs & MCP ExtensibilityWebhook style triggers plus agent deployment and monetization are documented; no SDK or MCP surface in retrieved materials, Twin materials 2026-07-22	Partial
Testing, Debugging & OptimizationLive feedback adjusts runs in progress and the vendor cites self healing reliability across three technology generations, short of a structured testing framework, Twin materials 2026-07-22	Partial
Browser & Computer UseThe proprietary Browser agent uses websites like a person, logging into legacy ERPs, vendor portals, and government sites, proven at scale by the Qonto Invoice Operator and OpenAI CUA alpha partnership, Twin and Tallyfy materials 2026-07-22	Full

Pricing

Credit based plans + custom enterprise

credits (Build Mode + Run Mode)

Public — partialMedium variable cost

What is public

The structure is public (credit based plans, Build Mode versus Run Mode credits, extra credit purchases, custom enterprise tier); specific plan rates were not captured in retrieved materials.

Variable cost rationale

Credit consumption scales with agent build and run activity; the vendor answers what typical tasks cost and when credits are charged, indicating usage driven economics.

Sales call required

No — self-serve available

Free / trial

Not captured in retrieved materials

Verified 2026-07-22

Data confidence: high