TL;DR:
- Twin Labs, a Paris-based startup, aims to automate repetitive business tasks.
- They utilize multimodal models, such as GPT-4V, to replicate human actions.
- Traditional language models were found to be unreliable for their purposes.
- Twin Labs operates like a web browser, handling tasks like data entry and clicking.
- The company plans to enable AI assistants to learn tasks through screen recordings and natural language.
- They have secured $3 million in pre-seed funding from prominent investors.
- Twin Labs intends to launch with pre-trained tasks and later allow clients to create customized tasks.
- Their approach focuses on streamlining daily tasks rather than chatbot-style interactions.
Main AI News:
Twin Labs, a dynamic Paris-based startup, is poised to revolutionize the world of business automation. Their mission? To simplify and streamline repetitive tasks that bog down modern workplaces. Whether it’s the intricate process of onboarding new employees, managing inventory levels, or gathering financial reports from multiple SaaS platforms, Twin Labs is set to transform the way businesses operate.
Founder and CEO, Hugo Mercier, explained the genesis of Twin Labs’ groundbreaking approach, stating, “Twin’s starting point is a science-fiction idea. We saw the development of the technical capabilities of LLMs — foundation models. And the question we asked ourselves was whether we’d be able to duplicate ourselves by training an AI agent on the way we perform our tasks.”
What sets Twin Labs apart is not just their objective but their innovative methodology. Instead of relying on traditional language models (LLMs), the company harnesses the power of multimodal models with vision capabilities, such as GPT-4 with Vision (GPT-4V), to replicate human actions.
Initially, Twin Labs experimented with autonomous agents using conventional LLMs but found them unreliable. Mercier noted, “Overall, the conclusion is that LLMs are completely unreliable. This means that LLMs are making the wrong decisions. In the end, the task isn’t done.”
GPT-4V, on the other hand, has undergone extensive training on various software interfaces and underlying code bases, allowing it to comprehend the functionality behind buttons and screens. Unlike traditional automation tools like Zapier, Twin Labs operates more like a sophisticated web browser. It effortlessly loads web pages, clicks on buttons, and inputs text, mimicking human actions seamlessly.
For example, when hiring a new employee, Twin Labs can handle a series of tasks, including adding the person’s information to the payroll system, sending invitations via Slack, creating a Google Workspace account, and facilitating account creation with the healthcare insurance provider. These tasks may not be inherently complex, but executing them accurately, in the right sequence, with specific criteria met, is crucial. Twin Labs is working towards enabling AI assistants to learn through screen recordings and natural language descriptions, ensuring tasks are performed precisely.
Twin Labs has been making significant strides toward its vision. Co-founders Hugo Mercier and Joao Justi have spent the past six months developing a prototype of their product. They’ve also secured $3 million in pre-seed funding from esteemed investors, including Betaworks, Motier Ventures, and a roster of influential angel investors.
While Twin Labs acknowledges challenges, such as the cost associated with task completion, the company is optimistic about the evolving landscape of AI infrastructure and APIs. Initially, they plan to launch a product with a library of pre-trained tasks to ensure optimal performance. Subsequently, Twin Labs intends to open up its platform, allowing clients to customize and create their own tasks.
In a domain where AI products often manifest as chatbot interfaces, Twin Labs stands out for its innovative approach to interacting with AI models. Hugo Mercier emphasized their mission, stating, “We really wanted to get down to the nitty-gritty of what people do on a day-to-day basis, and how we can take over some of the things that are actually a bit of a hassle for them.” Twin Labs is undoubtedly a trailblazer in the realm of business automation, promising a future where repetitive tasks are handled with efficiency and precision by AI-driven agents.
Conclusion:
Twin Labs’ innovative approach to business automation, using multimodal models and mimicking human actions, could disrupt the market by providing a more efficient and reliable solution for repetitive tasks. This approach may lead to increased productivity and reduced reliance on complex API-driven systems, making it an appealing option for businesses looking to streamline their operations.
