Just a few months ago, AI-powered automation agents were more of a futuristic idea than a practical tool. But now, Paris-based startup Twin is proving that the future is arriving faster than expected. The company, which emerged from stealth in January 2024, has announced a partnership with Qonto, a leading fintech provider of business banking solutions in Europe. This collaboration brings Invoice Operator, an AI-driven automation agent, into real-world use, streamlining invoice retrieval for businesses and reducing manual work significantly.
A Smarter Approach to Automation
Businesses have been automating tasks for years using tools like Zapier, which connects different apps through APIs, or robotic process automation (RPA) solutions like UiPath, which mimic user interactions to complete repetitive tasks. However, these traditional methods have limitations. API-based automation requires companies to integrate with each service individually, which can be time-consuming and resource-intensive. RPA, on the other hand, often breaks when websites update their interfaces, requiring constant maintenance.
Twin, with its compact yet ambitious nine-person team, is taking a different approach, leveraging artificial intelligence and advanced computer-use models to handle automation in a much more flexible and efficient way. Instead of relying on pre-built integrations or rigid scripts, Twin’s AI-powered agent can dynamically navigate websites, extract information, and complete tasks in a way that mimics how a human would interact with a browser.
How Does an Invoice Operator Work?
Qonto, which processes millions of invoices monthly, identified a common problem among its users, the time-consuming task of manually retrieving and uploading invoices. Twin’s Invoice Operator aims to solve this by automating the entire process.
Here’s how it works:
- The AI first scans the user’s transactions to identify missing invoices.
- It then determines which services it needs to access to download the invoices.
- A browser window appears, showing the agent performing actions in real time.
- If a login is required, the process briefly pauses, allowing users to enter their credentials manually.
- Once authenticated, the agent resumes, finds the necessary invoices, downloads them, and attaches them to the correct transactions in the user’s Qonto account.
This process, which could otherwise take hours of manual effort, is now handled efficiently with minimal user intervention.
Related links you may find interesting
Why Does Twin’s Approach Stands Out?
What makes Twin’s AI automation unique is its ability to adapt and scale without the need for extensive custom scripting. Traditional RPA tools require a separate script for each website, which means any small change to a website’s layout can break the automation. As CEO Hugo Mercier points out, that approach simply wouldn’t work at Qonto’s scale, where users interact with thousands of different service providers for invoicing.
Moreover, compared to API-based automation tools like Zapier, which took a decade to support 8,000 applications, Twin has managed to support thousands of services in just a few months. This speed of deployment is a testament to the power of AI-driven automation.
Behind the scenes, Twin operates using a Chromium-based browser running on a cloud server, powered by OpenAI’s Computer-Using Agent (CUA) model. Twin was among the first 15 companies to gain access to this cutting-edge AI technology during its beta phase, giving them an early advantage in refining their automation agent.
Making AI Automation User-Friendly
One of Twin’s biggest priorities has been making automation simple and accessible for businesses, even those that may not be tech-savvy. Unlike many AI tools that require users to write prompts or configure workflows, Twin’s approach is fully intuitive. Users just log into their accounts and launch the agent, which then navigates and retrieves invoices automatically.
According to Mercier, this emphasis on ease of use is what will set Twin apart in the growing AI automation market. Rather than targeting only technical teams or IT professionals, Twin is building solutions for everyday business users who want seamless automation without a learning curve.
The Bigger Vision: AI Agents Beyond Invoicing
While Invoice Operator is just the beginning, Twin has its sights set on expanding AI automation to other industries. The company envisions a future where AI agents can handle a wide range of tasks, such as:
- E-commerce automation:- AI agents could manage online orders, track shipments, or handle customer service inquiries.
- Marketplace management:- Businesses could automate cataloging, price comparisons, and product updates across multiple platforms.
- Call center support:- AI-driven agents could retrieve customer information, summarize interactions, and assist human agents in resolving issues faster.
Ultimately, Twin is betting on a future where AI agents are not just faster and more cost-effective but also highly adaptable across different business sectors.
The Future of AI-Driven Automation
The emergence of Twin reflects a broader shift in the automation industry. Companies are moving away from rigid, rule-based automation toward intelligent agents that can adapt to different workflows dynamically. The demand for AI-driven solutions is only growing, and if Twin continues on its current trajectory, it could become a major force in the AI automation landscape.
As the technology matures, the key challenge will be reliability and security. Businesses will want assurance that AI agents can handle tasks accurately and safely, especially when dealing with sensitive financial or customer data. If Twin can address these concerns while maintaining its speed, efficiency, and ease of use, it has the potential to transform the way businesses approach automation in the AI era.
For now, its partnership with Qonto is a strong first step and a glimpse into what the future of AI-driven automation could look like.
