Telex is a collaboration workspace where modern organizations pair human insight with intelligent agents to move faster, stay aligned, and execute at scale. It is built for fluid communication like Slack, but also adds AI coworkers into the conversation that don’t simply answer questions but take action. It's just like Slack, but for education and bootcamps.
In this article, I will show you how to integrate your AI agents with the Telex platform. This article is especially for those who are not using Mastra for agent deployment. I assume you have developed and deployed your AI agent already. If you haven't go ahead and do so now. This article talks about the specific AI agent I developed for Telex so that Java developers can have a guide on doing so (You can view the AI agent itself on Telex here).
For now, let's see how to integrate with Telex.
Step 1: Create an AI co-Worker
Head over to your Telex dashboard. You would notice a number of things, one of which is the sidebar on the left of the screen. Under the list of AI colleagues, you will see the option to add a new colleague. Click on it.
Next, fill the form that shows up (check the picture above). Ensure to set the visibility to public.
Step 2: Configure the co-worker's workflow
After creating it, you will see the name of the worker you just created at the top of the chat screen, and a chat screen. Click on the name of the worker, and on the sidebar that appears on the right, select "Configure Workflow".
Click "Add a task". Change the view to JSON. You will be using a JSON object to configure the workflow of your co-worker. Your co-worker makes use of your AI agent to carry out its task. This is a template workflow JSON that you can follow:
{
"active": true,
"category": "utilities",
"description": "An agent that explains what type of string a regex pattern matches",
"id": "c65ca3d5b711",
"long_description": "Your job is to explain, in very simple terms, the type of string that a regex pattern matches. You are to ask for a regular expression from the user. When a user sends a regular expression or asks what a particular regular expression does, provide the explanation using the node provided. If the string passed to you contains anything aside from a regular expression, extract only the regular expression. Use only the regular expression from the most recent text sent by the user. Use the regexplain tool to get the string that matches the pattern. Do not respond by yourself, use the node provided. If you are unable to use the node, then say you can't provide an explanation right now. Don't keep message history",
"name": "bituan_agent",
"nodes": [
{
"id": "regexplain",
"name": "Regex Explainer",
"type": "a2a/custom-a2a-node",
"typeVersion": 1,
"position": [
800,
-100
],
"url": "https://regexplain-production.up.railway.app/a2a/agent/regexplain"
}
],
"pinData": {},
"settings": {
"executionOrder": "v1"
},
"short_description": "take a regex pattern and return a short description of the string it matches"
}
Paste the JSON object in the field provided and change some of the properties to match your agent. The category, descriptions, id, name, node-id, node-name, node-url are all properties you should change.
Step 3: Testing and logs
Save the workflow and go test it out in the chat section. Check the logs for more information on what's going on in the background. The logs for a specific co-worker can be found at https://api.telex.im/agent-logs/{your-coworker-uuid}.txt
To get your coworker UUID, send at least one message to your co-worker. Look at the URL of your co-worker, you will see something like this:
https://telex.im/telex-ai-intergration/home/colleagues/{UUID-1}/{UUID-2}. The first UUID on the URL (UUID-1) is your co-worker UUID.
With that, your co-worker should be up and running, and it should make requests to your AI agent, and receiving appropriate responses. This interaction can be seen in the logs. Now you have all you need to integrate you AI agent with Telex. Go out there and build!
That's all for now. See you in my next post!
Top comments (0)