Build a Twitter (X) AI Agent (n8n + Mention)

Creating an AI-powered social media agent that monitors Twitter for specific keywords and replies automatically can help your business engage with potential leads and customers. This article walks you through building such an agent using n8n automation platform, Mention.com for social listening, and the Twitter API. The agent evaluates tweets mentioning relevant keywords, determines if a reply is appropriate, and crafts helpful responses with AI, all while promoting your business subtly.

If you're interested in automating your social media engagement and want access to the full n8n automation template and prompts used here, consider joining our free community AI Automation Mastery. There, you can download everything you need to replicate this setup and learn how to build AI-powered automations.

How the AI Twitter Agent Works

The core idea is to monitor Twitter for tweets containing specific keywords related to your business or niche. The agent then uses AI to:

Filter out tweets that aren't relevant or worth replying to
Write personalized, helpful replies that guide users toward your business resources
Post the replies directly to Twitter as responses to the original tweets

This process helps you engage with potential customers actively asking questions or seeking solutions, without manually scanning Twitter or risking spammy interactions.

Setting Up Keyword Monitoring with Mention.com

The first step is to establish a social listening alert on Mention.com that tracks Twitter for keywords relevant to your business. This alert feeds all matching tweets into a dedicated Slack channel that your n8n workflow monitors.

To set this up:

Log into your Mention.com account and go to the Feed tab.
Click Add a New Alert.
Select the Standard Alert option. This option balances ease of use and filtering flexibility.
Define your keywords carefully. For example, if you offer AI tools, you might track phrases like "Is there an AI tool" or "Best AI tool for". Avoid overly broad terms to reduce noise.
Use advanced query options to exclude unwanted terms. For instance, you can exclude "free AI tools" if you want to focus on other searches.
Monitor the live feed on the right side to validate your query and adjust it to reduce irrelevant mentions.
Choose Twitter as the only monitored platform for this use case.
Set language filters, such as English, to narrow results.
Connect the alert to a Slack channel by adding Slack integration at the bottom of the alert setup page. Follow the prompts to authorize and select your channel.
Set filters like influencer score to control which mentions appear. For example, setting a low threshold ensures you capture tweets from smaller accounts to maximize engagement opportunities.

Once complete, Mention.com will continuously push tweets matching your criteria into your Slack channel. This real-time feed serves as the trigger for your AI agent.

Configuring the n8n Automation Workflow

With the social listening feed in Slack ready, the next step is building the automation in n8n. This workflow listens for new messages in the Slack channel and processes each incoming tweet mention.

Slack Message Trigger

The workflow starts with a Slack trigger node configured to activate when a new message is posted to your monitored Slack channel. This ensures the agent reacts instantly to relevant tweets.

Filtering Noise and Avoiding Loops

To prevent wasting AI credits and avoid spamming, the workflow applies three key filters:

Only Twitter Mentions: It checks that the mention originates from Twitter, ignoring mentions from other platforms like Reddit or Facebook.
Exclude Self-Mentions: It filters out tweets posted by your own account to avoid replying to yourself or creating reply loops.
Exclude Retweets: Retweets are ignored since replying to them can waste resources and appear spammy.

Extracting the Tweet ID

Each Slack message contains a JSON payload with metadata about the tweet. Using an AI prompt with a light model like GPT-4 Mini, the workflow extracts the Tweet ID from the URL embedded in the payload. This ID is essential for fetching full tweet details and posting replies.

Fetching Tweet Content Without Using Twitter API Quota

Instead of using the Twitter API directly—which has strict limits—the workflow uses a hidden syndication API endpoint to get the full text of the tweet. This method helps conserve API calls while obtaining the necessary tweet content for analysis.

Evaluating Tweet Relevance with AI

The workflow then sends the tweet content to an AI model with a custom prompt. The AI acts as an expert community manager, deciding whether the tweet is a good candidate for a reply that promotes your business.

The prompt includes instructions to identify tweets where the user expresses intent to find an AI tool matching their use case, such as:

"Is there a good AI tool for video editing?"
"What's the best AI tool to build my resume?"

Non-questions or irrelevant tweets are automatically excluded, reducing spam and focusing on meaningful engagement.

Creating Helpful Tweet Replies

Once the AI determines a tweet is worth replying to, the workflow proceeds to draft a response. This process uses your business’s AI tools directory as a knowledge base.

Fetching Relevant Categories

The system queries your API to retrieve a list of categories from your AI tools directory, which currently contains over 300 categories. This data is formatted into readable text snippets to provide context for the AI when crafting replies.

Writing the Reply Tweet

The AI receives the original tweet and the formatted category content. Using a detailed prompt with examples of past successful replies, it generates a personalized tweet that:

Answers the user's question
Includes a relevant link to a category page on your website
Maintains a friendly and helpful tone

Examples in the prompt help the AI understand what kind of replies work best and what to avoid, such as overly generic or spammy responses.

Delaying and Posting Replies

To avoid looking like a bot or triggering spam filters, the workflow inserts a short delay before posting the reply. Then, it uses the Twitter API node in n8n to post the reply tweet referencing the original Tweet ID.

The OAuth credentials for your Twitter account must be connected in n8n to authorize posting.

Internal Notifications and Monitoring

After posting a reply, the workflow sends a message back to Slack showing the reply content and the AI’s reasoning behind it. This feedback loop helps you monitor the agent’s decisions and refine prompts if needed.

Additionally, the workflow adds a reaction emoji to the Slack message to indicate whether a reply was posted or skipped.

Benefits of an AI Twitter Agent

Running this agent 24/7 lets you:

Engage automatically with users asking questions relevant to your business
Drive traffic to your website or directory by linking relevant resources
Save time by automating social media management tasks
Maintain control and oversight through Slack notifications and reasoning logs

Because the agent filters tweets carefully and crafts custom-tailored replies, it avoids spammy behavior that could damage your social media presence.

Getting Started: Join the AI Automation Mastery Community

To help you build this AI Twitter agent yourself, we provide the complete n8n automation template, all prompts, and setup instructions for free. Join our AI Automation Mastery community to access these resources and connect with other AI builders.