How do I answer customer DMs with AI?

Q: How do I answer customer DMs with AI?

In the garden center case published by The SmallOffice Brief, Smith's Gardentown in Wichita Falls, Texas, uses one of its Grok Bots to comb every business social account. The bot finds the comments and direct messages that need a response. At 8 AM it sends the owner a daily report with a draft reply for each one. He scans the drafts, approves or edits them, and then the bot sends them. The bot also scans during the day for anything urgent. The owner reports the daily report and drafts are in place and calls the setup a game changer; the post gives no count of messages found or answered. That report is the source's, not a measurement by The SmallOffice Brief.

Answered from one published case only. Every claim below links to its case page. Updated Mon, Nov 2, 2026, from the cases this answer is based on.

Short answer: let a bot read every account for you. It finds each comment and direct message that needs a reply. It drafts a reply for each one. You approve or edit each draft. Then the bot sends it. That is the setup at Smith's Gardentown, a garden center in Wichita Falls, Texas, in 600,000 followers, every comment found daily.

The steps, from the case

Let the bot read every account. One of the owner's Grok Bots combs every business social account. The shop's accounts hold more than 600,000 followers in total, per the case. (case)
Let it find what needs a reply. The bot finds the comments and direct messages that need a response. No message waits unseen in an inbox. (case)
Get one report each morning. At 8 AM the bot sends a daily report. The report holds a draft reply for each message. (case)
Approve or edit each draft. The owner scans the drafts. He approves a draft or edits it. Then the bot sends it. Nothing sends before he approves it. (case)
Let it watch for urgent items. The bot scans again during the day. It flags anything urgent at once. (case)

What the source reports it achieved

The owner reports the daily report and drafts are in place, and he calls the setup a game changer. The post gives no count of messages found or answered, so no volume or time figure can be claimed from it. That is the owner's report in Michaelfiore on X, not our measurement. Read the full case: 600,000 followers, every comment found daily. The case page carries a copy-ready prompt based on the source's workflow, with its source link.

The same draft-and-approve job on reviews and guest messages

This case is not a DM case, and we do not present its results as DM results. It shows the same job on reviews, emails, and guest messages: the tool drafts each reply, and a person reviews and sends.

Case	Outcome, as reported	Source
Review replies: 30 minutes to five	Reply work fell from a 30-minute task to five minutes, guest messages responded to rose 898%, and time to respond to guest feedback fell 85%, per SevenRooms, attributed to the group's team. The figures are the vendor's blog, with no business-specific limit stated.	SevenRooms blog

Method: this page is written only from cases published in The SmallOffice Brief. The DM claims come from one case, linked throughout; the related case is labelled as a different kind of message. Figures are each source's claims, not ours, and are not independently verified by us.

More direct answers: all Answers · The numbers behind all our cases: Stats

How do I answer customer DMs with AI?

The steps, from the case

What the source reports it achieved

The same draft-and-approve job on reviews and guest messages

5 use cases every morning