Guide
AI agent for social listening: find people asking for what you sell
Updated · by Rosh Perera
Some of your best leads are people describing your exact problem in public: 'is there a tool that does this?', 'how do you handle that?'. An AI agent can find those posts and draft a useful reply. It shouldn't post for you. Reddit's rules say to participate authentically in communities where you have a personal interest and not to spam (Reddit Rules). X's developer policy says to get explicit consent before sending people automated replies or Direct Messages (X Developer Policy). So the setup is simple: the agent listens and drafts, and you read, edit and post every reply yourself from your own account. No auto-posting, no auto-DMs and no extra accounts. Time figures here are estimates, not measurements.
What to hand to an agent first
Problem phrases, not brand names
Write down the phrases people use when they have the problem you solve, plus a short list of subreddits and X searches where those people ask. Google Alerts can also email you when new results for a phrase show up in Google Search (Google Search Help).
Find posts the allowed way
Use each platform's own search or official API, or paste in posts you find yourself. Reddit's User Agreement prohibits scraping without Reddit's prior written consent, and its Data API Terms require a separate agreement for commercial use (Reddit). X's Terms of Service prohibit crawling or scraping without X's prior written consent (X Terms of Service).
Shortlist by fit
For each post, the agent notes what the person asked, how close it is to what you sell, and whether that community allows product mentions. It skips weak fits instead of replying to everything.
Draft a reply that helps on its own
It drafts an answer that solves the question even if the person never buys. If mentioning your product fits and the community allows it, the draft says plainly that it's yours. The FTC's Endorsement Guides cover disclosing a material connection (16 CFR 255.5).
You post, then track
You read each draft, edit it and post it yourself, or skip it. The agent keeps a simple log of the posts you answered, the replies you got and any signups you can trace, so after two weeks you can decide whether it's worth keeping.
Not sure which jobs fit your business?
Build my free AI plan →Or get every routine pre-written in the pack →$29 one-time, plus local tax where it applies.
Ready-made use cases
- 01Find buyers asking for help on X
The agent searches X for people describing the exact problem you solve and drafts a helpful reply for each.
~5h/week · estimate - 02Never miss a lead follow-up
The agent checks your CRM or inbox for leads that went quiet and drafts the next follow-up.
~4h/week · estimate - 03Weekly competitor watch
The agent tracks competitors' pricing, launches and posts and tells you what matters.
~2h/week · estimate - 04Trend radar to content ideas
Each morning the agent scans your niche for trends and gives you 5 timely post angles.
~3h/week · estimate - 05Daily X posts from your real work
The agent drafts daily posts from what you actually did (commits, wins, calls) so you post consistently.
~4h/week · estimate
Example: Grok Bot setup
Grok Bot routine: Find buyers asking for help on X Schedule: Daily 10:00 Every day, search X for posts from the last 24h where people ask for help with [my service] or complain about [problem I solve]. Skip bots and competitors. Give me the top 10 with link, why they're a fit, and a helpful non-salesy reply draft. Never post without my OK. Rules: draft first, never send or post anything external without my approval. Keep summaries short.
Grok Bot, ChatGPT, Claude and n8n versions: all templates
Setup steps
- Write 10 problem phrases your buyers use, and pick 3 to 5 subreddits or X searches where they ask.
- Read each community's rules on self-promotion and note which ones allow product mentions.
- Have the agent check those places daily through the platform's own search or official API, and shortlist posts with a reason for each.
- Ask for one helpful draft reply per post, with your connection disclosed if it mentions your product.
- Edit and post every reply yourself from your own account, never by automation, and review the log after two weeks.
Keep these human
- Read each subreddit's rules before replying. Some limit or ban self-promotion, and Reddit's rules say to abide by community rules (Reddit Rules).
- On X, no automated replies or DMs without the person's explicit consent, no bulk or spammy actions, and no identical or near-identical posts across accounts (X Developer Policy).
- Use one real account, run by you. Reddit's rules say not to intentionally mislead others or impersonate anyone (Reddit Rules).
- Keep only what the log needs. Reddit's Data API Terms bar using its data to spam users and limit what you can keep (Reddit Data API Terms).
- Unlike most routines, this one never moves to auto-posting. You approve and post every reply, for good.
Want the whole system pre-written?
The One-Person AI Business Pack: CEO agent + workers, all 46 routines in 4 formats, your role pack.
$29 one-time, plus local tax where it applies.
Next: Whop checkout. $29 one-time, plus local tax where it applies, paid by card on Whop. Files arrive by email.
14-day money-back guarantee: reply to your receipt within 14 days for a full refund.
Built by Rosh Perera, who has set up AI automations for more than 40 businesses over about 3 years.
FAQ
Where should I start?
Problem phrases, not brand names: Write down the phrases people use when they have the problem you solve, plus a short list of subreddits and X searches where those people ask. Google Alerts can also email you when new results for a phrase show up in Google Search (Google Search Help). Find posts the allowed way: Use each platform's own search or official API, or paste in posts you find yourself. Reddit's User Agreement prohibits scraping without Reddit's prior written consent, and its Data API Terms require a separate agreement for commercial use (Reddit). X's Terms of Service prohibit crawling or scraping without X's prior written consent (X Terms of Service). Shortlist by fit: For each post, the agent notes what the person asked, how close it is to what you sell, and whether that community allows product mentions. It skips weak fits instead of replying to everything.
What should stay human?
Read each subreddit's rules before replying. Some limit or ban self-promotion, and Reddit's rules say to abide by community rules (Reddit Rules). On X, no automated replies or DMs without the person's explicit consent, no bulk or spammy actions, and no identical or near-identical posts across accounts (X Developer Policy). Use one real account, run by you. Reddit's rules say not to intentionally mislead others or impersonate anyone (Reddit Rules). Keep only what the log needs. Reddit's Data API Terms bar using its data to spam users and limit what you can keep (Reddit Data API Terms). Unlike most routines, this one never moves to auto-posting. You approve and post every reply, for good.
Is there a ready-made template?
Yes. Start with "Find buyers asking for help on X": copy-ready Grok Bot, ChatGPT, Claude and n8n templates are at https://whatshouldmybotdo.com/use-cases/lead-finder-on-x#templates.