I Tried The “Easiest” Ways To Make Money With AI

Executive summary

Codie Sanchez tested five AI-agent workflows over 10 days, with a stated budget cap of $100 and a rule that she would not do the work after setup. The strongest business use was automated, personalized cold outreach, which generated 10 interested replies and two booked calls.

The video values those outcomes at roughly $2,000, based on the company’s estimated acquisition costs. That figure represents pipeline or acquisition-equivalent value, not confirmed cash revenue.

The broader lesson: use AI to handle repeatable research and administrative work, but set clear limits and keep humans involved in spending, contracts, and sales conversations.

Playbooks and business applications

Resale operations

Three agents divided the workflow:

• Find free or underpriced listings.
• Check shipping and resale economics, rejecting items that fail the margin test.
• Coordinate pickup, relist the item, and handle buyer inquiries.

The experiment reported $130 in sales, about $40.98 in cost of goods, and $1 in bot costs, for roughly $88.02 in net profit. One bowling ball was found, picked up, and delivered in 31 minutes. The process was also applied to unwanted office items.

Supplier purchasing and negotiation

The same multi-agent workflow could be applied to business procurement:

1. Set a target price and a hard limit the agent cannot exceed.
2. Request quotes from multiple suppliers.
3. Compare total costs, including shipping and minimum-order requirements.
4. Negotiate using the best competing offer.
5. Have a person review and approve the final purchase.

Sanchez gave a hypothetical example: on $200,000 of annual parts spending, an 8% discount would save $16,000 per year. This was a projection, not a measured result.

Content clipping

The bot searched for freelance work and produced short-video edits. It found an Upwork request for five clips at $40, but the client did not convert, so external revenue was $0. The clips were rated 4/10 against the creator’s other videos, and commenters did not identify them as AI-made.

The practical benefit was using the bot for editors’ first drafts and highlight selection. The video estimates potential savings of $50–$250 per video. At one video per week and a $150 average saving, that would be about $600 per month; the higher estimate was about $1,000 per month. The reported test costs appear to total roughly $31.76, though the subtitles’ financial figures are garbled.

Lead reactivation and email marketing

An agent built a landing page and personalized emails for 3,000 past book-launch contacts, then prepared the campaign in Klaviyo. Initial drafts needed feedback. With direction, the bot improved the page and copy and incorporated public customer reviews.

In a two-hour test, the campaign generated two book sales and $55.48 in revenue. The stated costs were $24 for goods and about $17 for tools, leaving roughly $14.48 net.

The reusable lesson is to revisit existing leads and customers: they are often cheaper to convert than newly acquired prospects.

Cold-outreach funnel

A lead-research agent gathered business owners’ contact details from sources such as Google and LinkedIn, with Apollo and Reddit also mentioned. Before outreach, it checked HubSpot to avoid contacting people already being handled by the team.

The system then:

• Verified email addresses with Hunter.io to reduce bounces.
• Created location-specific versions of cold emails.
• Started at 50 emails per day per mailbox, then expanded to four mailboxes and 200 per day.
• Routed replies to human sales managers for follow-up.

The stated target was local business owners with at least $1 million in revenue, for an Austin workshop. The test produced 10 interested responses and two booked calls. Using the company’s benchmarks of $100 per response and $500 per booked call, the video assigns the activity about $2,000 in acquisition-equivalent value.

The video reports costs inconsistently—about $57.26 in one calculation and roughly $187 later—so the claimed net figure should be treated cautiously, not as verified profit.

Trading bot

A bot monitored Reddit, X, and news, then traded through Robinhood. It lost money and was shut down. Sanchez’s takeaway was that automation cannot compensate for a poor strategy: a bad algorithm can simply lose money faster. She cautioned against using social-media hype as a trading signal.

Management takeaways

• Start with a specific workflow and measurable outcome, rather than deploying a general-purpose bot.
• Set hard spending and negotiation limits; require human approval for purchases and contracts.
• Use AI first inside your own operation to test quality and save time before trying to sell the service to customers.
• Measure real revenue, cost savings, and conversion outcomes separately. In particular, distinguish booked-call pipeline value from collected cash.
• For outreach, protect deliverability through email verification, controlled sending volumes, personalization, and suppression of leads already being contacted.
• Prioritize unworked leads, abandoned opportunities, and routine operational tasks. In this test, those workflows outperformed the more attention-grabbing trading and resale experiments.

Presenter: Codie Sanchez Platforms and sources referenced: Grokbot, Facebook Marketplace, Upwork, Robinhood, Shopify, Klaviyo, HubSpot, Hunter.io, Google, LinkedIn, Apollo, Reddit, X, and Amazon customer reviews.