The rise of the AI digital worker has unlocked unprecedented levels of productivity. Autonomous agents like Lexi can conduct market research, analyze complex datasets, and draft reports, freeing up human teams to focus on strategy and innovation. This is a massive leap forward in workflow automation.
But what happens when a task is unique to your business? What about fetching data from your proprietary CRM, updating a project in your specific task management software, or interacting with a custom-built internal API?
This is where standard AI assistants hit a wall. A truly effective digital coworker needs to do more than just access the public internet; it needs to learn the tools-of-the-trade specific to your company. This is where Lexi.do transforms from a powerful tool into a truly integral team member. It's time to move beyond simple delegation and start teaching your AI new skills.
Generic automation is powerful, but it often only gets you 80% of the way. The final 20%—the most valuable part—lies in your company's unique processes, proprietary data, and custom software stack. These are your competitive advantages.
An autonomous agent that can't interact with these systems is like a brilliant new hire who hasn't been given access to the company's internal tools. They can do great work in a vacuum, but they can't fully integrate into the workflow.
To achieve true end-to-end automation, your digital worker needs the ability to learn and use the very same tools your human employees do. It needs to be extensible.
With Lexi.do, you can define "Custom Capabilities"—essentially, teaching Lexi new skills. You can equip your digital worker with secure access to your internal tools, APIs, and databases.
Think of a capability as a single, well-defined tool that you add to Lexi's toolbox. Just as you might teach a human employee how to use Jira, Zendesk, or your internal billing system, you can teach Lexi to do the same.
Here are a few examples of what you could teach Lexi:
Once a capability is defined, Lexi can intelligently decide when and how to use it to accomplish a more complex objective you've delegated. This transforms Lexi from a generic agent into a bespoke digital employee, molded perfectly to your operational needs.
Teaching Lexi is an intuitive process for any developer, embodying the principle of Business-as-Code. You define your business's unique operations as functions that your AI agent can execute.
Let's walk through a practical example: teaching Lexi how to create a support ticket in your company's project management system.
import { Lexi, Capability } from '@do/sdk';
// 1. Initialize Lexi with your API key
const lexi = new Lexi({ apiKey: 'YOUR_API_KEY' });
// 2. Define the new capability
const createSupportTicketCapability: Capability = {
name: "createSupportTicket",
description: "Creates a new customer support ticket in the internal system. Use this for bug reports, feature requests, or customer issues.",
parameters: {
type: "object",
properties: {
title: { type: "string", description: "The title of the support ticket." },
description: { type: "string", description: "A detailed description of the issue." },
priority: { type: "string", enum: ["High", "Medium", "Low"], description: "The priority level of the ticket." }
},
required: ["title", "description", "priority"],
},
// This function contains the actual logic to call your internal API
// It runs securely in your own environment.
action: async ({ title, description, priority }) => {
// Replace with your actual API call to Jira, Zendesk, etc.
const response = await fetch('https://api.your-company.com/tickets', {
method: "POST",
headers: { 'Authorization': `Bearer ${process.env.YOUR_INTERNAL_API_KEY}` },
body: JSON.stringify({ title, description, priority })
});
const result = await response.json();
return `Successfully created ticket with ID: ${result.id}`;
}
};
// 3. Register the new capability with Lexi
await lexi.registerCapability(createSupportTicketCapability);
console.log("Lexi has learned the 'createSupportTicket' skill!");
// 4. Now, delegate a task that requires this new skill
const task = await lexi.delegateTask({
objective: "A customer reported a bug where the login button is not working on the checkout page. Create a high-priority support ticket for the engineering team to investigate.",
context: "Customer email: user@example.com. The issue is blocking purchases."
});
console.log(`Task delegated. Lexi will now use its new skill to create the ticket. Track status at: ${task.statusUrl}`);
In this example, Lexi receives a high-level objective in plain English. It understands the user's intent, recognizes that creating a "support ticket" is the required action, and autonomously uses the createSupportTicket capability you provided, filling in the title, description, and priority from the context of the prompt.
The ability to create custom capabilities is what separates a simple AI tool from a true AI Digital Worker platform. It allows you to:
Stop trying to fit your complex business processes into the rigid boxes of pre-built tools. Instead, build an autonomous agent that learns, adapts, and operates within the unique landscape of your organization.
Ready to build your custom autonomous AI workforce? Explore the Lexi.do documentation or request a demo to see custom capabilities in action.