Orchestrate every channel
from one intelligent
layer.
In 30 minutes we'll show you how Twinny connects to your operation and starts delivering results from week 1.
Your software has no API? We can automate it too.
Legacy applications, in-house software, third-party portals and web applications with no integration available. We combine AI, RPA and browser automation so a digital worker does on screen what a person would do.
Definition
Who it is for
- CIO / CTO
- Operations
- Administration
- IT
Automation without an API. Automation without an API means operating a system through its user interface (screens, forms, desktop or web applications) when no API or connector is available. Twinny combines three techniques: artificial intelligence to understand the request and decide, RPA to execute reliably in desktop and legacy applications, and AI-driven browser automation for web applications that change frequently. The result is wrapped as a reusable capability (MCP) for any digital worker.
When automation without an API is needed
- Old ERPs or management systems (AS/400, client-server applications, Citrix)
- In-house software built years ago with no integration layer
- Supplier, government or insurer portals that only offer web access
- SaaS applications whose API does not cover the operation you need
- Closed academic, clinical or booking systems
How we do it: AI + RPA + browser automation
- AI understands and decides The digital worker interprets the request, looks up the Company Memory and determines what must be done in the system.
- RPA executes in legacy applications Robots that operate desktop applications and terminals reliably, with validations and retries.
- Browser agents for web applications AI-guided browser automation that adapts to screen changes instead of breaking.
- Everything is wrapped as a capability The operation is exposed as an MCP service reusable by other agents, with permissions and audit.
What changes compared with traditional RPA
| Traditional RPA | Twinny: AI + RPA + browser | |
|---|---|---|
| When a screen changes | It breaks | The agent adapts within the rules |
| Unstructured inputs (email, PDF, photo) | Not understood | IDP and language understanding |
| Decisions | Fixed rules | Reasoning with company context |
| Reuse | One robot per process | MCP capability reusable by any agent |
| Exceptions | Failure | Escalation to a person with the case prepared |
Security and traceability, also without an API
Operating through the interface does not mean operating without control. Every action by the robot or browser agent runs with an identity, within a scope, with validations before confirming and with a log of what was read and written. Critical actions go through human approval.
Related cases
Industrial −72% cycle time per invoice ROI 8.5x · €210,000 saved/year
Supplier invoices: from email to the ERP without passing through anyone
Read case
Education −78% processing time per record ROI 7.2x · €140,000 saved/year
From the PDF record to the academic system, with no manual processing
Read caseFrequently asked questions
- Can software without an API be automated?
- Yes. It is operated through its interface with RPA (desktop and legacy applications) or with AI-guided browser agents (web applications), wrapping the operation as a reusable capability.
- What happens when the screen changes?
- Traditional RPA breaks. AI browser agents interpret the interface and adapt within the defined rules; if something is ambiguous, they escalate to a person.
- Is it secure?
- Every action runs with its own identity and scope, with prior validations, full logging and human approval for critical actions.
- Does it work to integrate a legacy system with AI agents?
- Yes. It is the usual way to connect old systems to the AI Operating Layer: the interface operation is exposed as an MCP capability any authorised agent can invoke.
Start with one process.
Tell us what you want to automate and you will receive an initial proposal within 24 hours.