Real AI agent use cases, with sources and setup prompts

Run an AI SDR team with human approval

Daily, three Sonnet agents take 15 leads each, research the site and Meta Ads Library, find weaknesses and write a cold email; Opus reviews and he approves.

Free to start3-5 hoursAdvancedSource linked

Source

@AI_Bhaiiii on X · 2026-09-21

Written in our own words. Credit to the original author. Link checked 2026-10-14.

What they did

Daily, three Sonnet agents take 15 leads each, research the site and Meta Ads Library, find weaknesses and write a cold email; Opus reviews and he approves.

Result (as claimed)

Presented as his daily routine; reply rates not given.

Tools they used

Claude CodeClaude SonnetClaude OpusMeta Ad Library

Tools in our setup

Suggested in the setup steps below; not named in the source.

HubSpotApollo.ioGrok BotPerplexitySuperhuman

Cost: Apollo Free plan, HubSpot Free CRM (up to 2 users), Clay not required; Grok Bot or n8n for the agent loop.

Set it up

  1. Define your ideal customer

    Write my ideal customer profile: industry, size, roles, trigger events, and 3 disqualifiers. Then turn it into Apollo search filters.

    Docs: docs.apollo.io/

  2. Pull a small list

    Pull 25 contacts that match the filters from Apollo using APOLLO_API_KEY, skip anyone without a verified email, and save them to a CSV for review.

    Docs: docs.apollo.io/reference/people-api-search

  3. Research each account

    For each company, find one recent, specific, public fact (launch, hire, post) with a source link. If you cannot find one, say so.

  4. Draft, don't send

    Draft a 4-sentence email per contact using the fact, a clear reason for reaching out, and an easy opt-out line. Put drafts in HubSpot as tasks for my approval; do not send anything.

    Docs: developers.hubspot.com/docs/api/crm/contacts

  5. Log in the CRM

    Create or update each contact in HubSpot with HUBSPOT_ACCESS_TOKEN (private app with crm.objects.contacts scopes) and log the drafted email and source.

    Docs: developers.hubspot.com/docs/api/private-apps

Safety: No spam: small, relevant lists and a human approves every email. Follow CAN-SPAM: accurate headers and subject, your physical address, and a working opt-out honored within 10 business days. Do not scrape sites against their terms; use data providers and public sources.

Copy the full setup prompt

Paste into your agent and it will walk you through the setup, asking before anything that sends, posts or spends.

You are helping me set up this AI use case end to end in Claude Code.

Goal: Run an AI SDR team with human approval
What the original author did: Daily, three Sonnet agents take 15 leads each, research the site and Meta Ads Library, find weaknesses and write a cold email; Opus reviews and he approves.
Reported result (as claimed by the author): Presented as his daily routine; reply rates not given.
Source: https://x.com/AI_Bhaiiii/status/2102058314280710503
Tools they used: Claude Code, Claude Sonnet, Claude Opus, Meta Ad Library
Category: Marketing & growth

Please:
1. Ask me 3 short questions about my situation (accounts I have, budget, what I want the output to look like).
2. Propose a step-by-step plan using these tools or cheaper alternatives, with the time and cost of each step. Use official pricing pages; say "check pricing" if unsure.
3. Then do the setup with me one step at a time, waiting for my OK before anything that sends messages, posts publicly, spends money or deletes data.
4. Finish with a checklist to run it again next time.

Related use cases