Make, an enterprise-grade visual-first no-code platform for creating and automating app workflows, today announced the launch of Make AI Agents, which will use artificial intelligence to help users build and orchestrate processes that can adapt in real time.

The company provides an intuitive interface for teams to visualize and collaborate on automating application and data work using simple drag-and-drop. Users can set up triggers such as Gmail to begin when an email comes in to open a Google Sheet to add information to a row and then inform people in Slack.

A slightly more complex automation could be created by passing the email through OpenAI’s ChatGPT to have it provide a concise sentence summary of the email for Slack or adding a separate trigger to generate a ticket for someone on the marketing team if specific keywords appeared in the email mentioning particular products.

Make connects to more than 2,000 apps that can exchange data with automated conditional logic. Users can set them up using the visual interface already as easily as connecting them and using drop-downs and a simple setup.

Make AI Agents simplify this even further by using natural language prompts to understand the goal of the user and make context-aware decisions as they work.

AI agents are a type of autonomous AI software that can take actions independently or semi-autonomously and work towards a given goal. They are distinct from AI assistants in that they can work with little or no human interaction.

Since AI agents can break down that goal into steps and take action, the Make AI agent can respond to changing conditions in the workflow environment and access relevant tools to address current needs based on the apps it has in its toolbelt.

“With Make AI Agents, automation moves into a new realm of intelligence and flexibility,” said Fabian Veit, chief executive of Make. “When businesses can automate without rigid rules and setups, the capabilities are endless. Make AI Agents can be tailored and configured for different use cases, driving intelligent and efficient automated decision-making.”

Users begin by describing the agent’s role and behavior in a brief description, essentially defining its goal. Next, the user attaches tools that provide the agent with the necessary skills and applications it can access to achieve this goal. After that, the user activates the agent with a prompt, which can be sent via Slack, email, or any other platform where they expect it to listen. The agent will then use its reasoning abilities to handle the requests.

For example, users could tell the AI Agent they want it to act as an inventory assistant in their system and help manage stock. In its description, users tell it that they want it to be able to answer questions about what’s in stock, order new stock and look up current order status.

They then build tools for the agent to use, such as an action set that lists inventory, orders stock or looks up order status. Finally, users connect the agent to inputs and outputs, such as Slack so that they can talk to it.

Now users can query the agent in Slack to ask things such as “How many green rabbit figurines do we have in stock?” or “Order 50 more gray fox plush toys” to get the agent to respond.

According to Make, what sets agents apart from a more rigid set of tools and triggers is that agents are more flexible and easier to work with. Since they will adjust to changing conditions, they require less maintenance. If something changes in the workflow, they can adjust around it in most circumstances using AI reasoning to handle most challenges.

The company stated that users can select the third-party large language model that best fits their needs, including models from OpenAI.

Image: Make