Adept Unveils Advanced AI Agent with Multimodal Capabilities

Adept introduced a groundbreaking development. Artificial Intelligence The new AI agent is designed to simplify complex web interactions and enterprise automation using a custom Adept Workflow Language (AWL). According to Adept.ai, AWL is a powerful language that allows users to easily create sophisticated workflows.

Key features of Adept’s AI agent

Adept’s AI agents are designed to be reliable, robust, and user-friendly. They can translate user intent into actions, managing everything from complex tasks to repetitive household chores. The agents are built on a collection of multimodal models trained to understand the screen, reason about content, and plan from the very beginning of training.

Key features of Adept’s AI agent include:

• Trustworthy: Agents can continuously execute workflows while maintaining the “rails”.
• Robustness: They are resilient to changes in the execution environment and remain functional despite the changes.
• Ease of writing: Instructions for counselors are easy to write and can be written in a few lines in natural language.

Adept Workflow Language (AWL)

AWL, a proprietary language developed by Adept, is a syntactic subset of JavaScript ES6. It provides a powerful abstraction for defining multi-modal web interactions. Users write workflows in AWL, which are directly translated into model calls. Specific AWL functions allow users to write commands in natural language, which are then translated into detailed AWL by the model.

AWL has the following functions: click(“Compose”) and act() It provides a flexible way to define agent behavior. The former finds elements on the screen and interacts with them by making function calls, while the latter takes natural language input and invokes an agent inference loop, allowing the agent to plan and dynamically execute actions.

Real Applications

Adept demonstrated how to use AWL to create workflows for a variety of applications. One example was viewing a PDF of an event attendee, extracting information, and creating a new lead in Hubspot. The same workflow was replicated in Salesforce with minimal changes, demonstrating the flexibility and efficiency of AWL.

Another example highlighted extracting patient diagnostic notes from a PDF and entering them into an EMR system. Agents can also handle tasks such as creating customer records in Stripe from Google Sheets data and managing leads in Salesforce based on inbound emails.

Transformability

Adept’s AI agents have the potential to transform business operations. The ability to create automation in natural language lowers the barrier to adoption, enabling more users to become citizen developers. This can reduce costs, accelerate time to value, and increase the desire for automation across a wide range of workspaces.

Adept’s agents can also handle more complex workflows involving unstructured data, improving resilience and robustness. Inference-time inference allows them to adapt to underlying web changes, making them suitable for a wide range of applications.

Future outlook

Adept is committed to making automation easier for a wider audience, enabling them to perform more types of workflows more frequently. The potential for business transformation through Adept’s AI agents is significant, providing simpler, faster automation and the ability to automate more complex tasks.

For more detailed examples of workflows built using AWL, visit the Adept.ai blog.

Image source: Shutterstock