Using ChatGPT to make $100 cold-calling as a complete stranger

Executive overview

Dan Martell attempts to earn $100 in under an hour using only ChatGPT, with no existing audience, no contacts, and no money. He lets GPT drive every decision — who to call, what to offer, what to say.

The challenge surfaces every real obstacle a first-time founder faces: slow AI tools, unresponsive prospects, bad demos, and no social proof. The core insight is that speed of iteration matters more than a perfect offer.

• Cold outreach works best via text, not calls
• A working demo closes faster than any script
• AI video tools for real estate are an unsolved market gap

Rules and setup

• No existing audience, followers, or contacts — full anonymity
• ChatGPT must drive every decision; Dan can only prompt it
• $100 must be received before the 60-minute timer ends
• Alter ego created: "Jeff Mercer" — to avoid recognition

GPT-driven offer selection

• GPT suggested targeting local real estate agents — decision-makers who answer their phones
• Offer chosen: AI listing video package — $100 for three branded property videos
• GPT generated a Hormozi-style offer, a cold-call script, and objection responses
• Payment collection problem identified late: Dan is too recognisable locally

Cold calling results

• Three calls made; all answered — real estate agents are reliable phone pickers
• Common response: "I get three of these calls a day" — immediate credibility gap
• Every prospect asked for an example before committing
• Decision stalled on every call; no one closed verbally

Pivot to text outreach

• Switched from calls to SMS after recognising phone calls create decision delay
• Message: "Got two free video slots left this week. Got a listing to test?"
• Two responses within minutes — one prospect sent a listing to demo
• Text outreach proved faster and lower friction than cold calls

Demo creation attempt

• Fed MLS listing into ChatGPT o3 → generated video script → passed to Google Veo3 for video
• Veo3 output: wrong images, misspelled "vaulted ceilings," unusable for a client demo
• Tried three separate AI video tools — all failed to produce a usable listing video
• Root cause: no turnkey tool exists that ingests an MLS page and outputs a social-ready video

Market gap identified

• Every AI video tool tested required manual assembly — no automated pipeline
• Dan estimates that a tool solving this end-to-end could generate $100 every 10 minutes at scale
• GPT should have been used first to identify the right tool before committing to the offer

Final minutes and outcome

• Last three minutes: lifeline granted — allowed to use real name and call known contacts
• Called real estate agent friend AJ — no answer
• Timer expired with $0 earned; Dan hands over his car keys for the day
• Post-challenge reflection: two hours of planning + 30 minutes of calls got close to a real sale