How to set up X DM auto replies without turning your inbox into spam

Step 1: Decide who should be able to message you

Open X, then review Privacy and safety settings for Direct Messages before adding any automation. X currently lets accounts receive requests from everyone, verified users, or eligible subscribers depending on account settings. The official X Direct Message guide explains the available controls and how message requests differ from established conversations.

A wider inbox can create more opportunity and more noise. Record which request folders you will check, whether subscriber messages matter to the business, and what should happen when a message is spam, abusive, or unrelated. Automation should not convert an unwanted request into an ongoing conversation.

• Review message-request settings
• Keep block and report behavior intact
• Do not treat every follower as opted in
• Write down the inbox purpose

Step 2: List the five questions you answer repeatedly

Review a representative sample of inbound DMs and count repeated intents, not repeated words. “What do you post,” “what is on your page,” and “what do subscribers get” may all be the same routine question even though the wording differs.

Start with no more than five answerable intents. For a subscription creator, useful candidates may include what the creator shares, where the official destination is, whether new updates are posted, a basic availability question, and a follow-up asking for one approved detail. Keep payments, custom requests, collaborations, personal disclosures, disputes, and uncertainty outside the automatic set.

Step 3: Write the facts before writing the personality

For each routine intent, write the exact facts the assistant may use. Include the creator's approved description, official destination, stable availability details, and any claim that must never be made. If a fact changes weekly, give it an owner and a review date instead of assuming the model will know it.

Then add the voice layer: normal greeting, sentence length, emoji habits, favorite phrases, and wording to avoid. Voice changes presentation; it must never create a discount, promise, or offering that is missing from the approved facts. Use the DM voice training guide for a deeper profile worksheet.

• Approved offering description
• Official destination
• Stable availability facts
• Forbidden claims and phrases

Steps 4 and 5: Connect officially and begin with drafts

Authorize the X account through the product's official connection flow. Do not give an automation provider the creator's password, recovery code, browser cookie, or session export. Confirm the requested permissions and keep a documented disconnect path.

Run the first twenty real replies in approval mode. Test a routine question, a follow-up that depends on prior context, a brand inquiry, a payment concern, a custom request, and a vague message. A passing test gives the routine questions useful answers and sends the risky examples back to the creator without an automatic response.

Steps 6 and 7: Set limits, review the first week, then widen carefully

Set a hard monthly reply allowance before enabling automatic sending. Decide whether replies should pause at the limit or use a visible overage rate. A creator should never learn about unexpected usage after the invoice arrives.

Review every correction and handoff during the first week. If a reply sounds wrong, determine whether the missing input is a fact, voice rule, thread detail, or handoff condition. Fix that source instead of adding a longer generic prompt. Loresta starts with 20 approved replies free so this review can happen in the real workflow without a countdown.