I Gave ChatGPT, Claude, and Gemini the Same Cold Email Brief — Only One Actually Got Replies
Most AI cold email comparisons are useless. They pick a generic prompt, paste three outputs side by side, and call it a day — no context, no follow-up, no real-world testing. I did this differently: I used an actual outreach brief, sent variations to real prospects, and tracked open rates and replies over two weeks. The differences between these three AI tools weren't subtle — they were the kind of differences that mean the gap between zero replies and booked calls. By the end of this article, you'll know exactly which AI to use, how to prompt it, and why the "best writer" isn't always the best tool for this job.
The Brief I Gave All Three (And Why the Setup Matters)
The brief was simple and realistic — the kind any freelancer or founder might write. Here it is:
"Write a cold email to a SaaS founder whose company just raised a Series A. I'm a freelance conversion copywriter. The goal is to get a 20-minute call. Keep it under 150 words. Don't use buzzwords. Make it feel human."
That's it. No extra hand-holding. I wanted to see what each AI did with a clear but open-ended brief, because that's how most people actually prompt these tools.
ChatGPT (GPT-4o) came back fast with something confident and structured. It had a hook, a value proposition, and a soft CTA. It read like a copywriter wrote it — maybe a little too much like a copywriter wrote it. The phrases "drive conversions" and "unlock growth" showed up despite me explicitly saying no buzzwords.
Claude (claude-3.5 Sonnet) did something different. It asked a clarifying question first — which surprised me — then produced an email that felt almost eerily personal. It referenced the emotional reality of post-fundraise pressure without being manipulative. It sounded like a person, not a tool.
Gemini (1.5 Pro) gave me a solid, professional email. Clean structure, good length. But it felt like the AI was playing it safe — the kind of email that won't offend anyone and probably won't move anyone either. It was the safest output. Which, in cold email, is the most dangerous thing to be.
What the Actual Outputs Revealed About Each AI's Personality
Every AI has a writing personality, and cold email is one of the best stress tests for it because the stakes are real and the margin for fluff is zero.
ChatGPT writes to impress. It defaults to structured persuasion — hook, proof, CTA — because that's what it's seen rewarded in training data. When you give it a cold email brief, it reaches for what "looks like" a great cold email. The problem is that the best cold emails don't look like cold emails. GPT-4o needed two rounds of refinement before it stopped sounding like a sales template.
Claude writes to connect. This is its defining quality. When I gave it the same brief, it prioritized the reader's emotional state over the sender's pitch. The email it produced acknowledged that Series A founders are suddenly flooded with vendor outreach — and positioned my hypothetical copywriter as someone who understood that pressure. That's not just good writing. That's empathy-driven positioning, and it's incredibly hard to prompt into existence with the other tools.
Gemini writes to complete. It treats the task as a checkbox. The output is competent, grammatically perfect, and forgettable. Gemini 1.5 Pro is genuinely powerful for research, summarization, and long-document work — but for persuasive short-form copy that needs to feel human, it's currently the weakest of the three.
The deeper insight here is that you're not just choosing an AI — you're choosing a default worldview. ChatGPT defaults to persuasion. Claude defaults to empathy. Gemini defaults to completion. Knowing this changes how you prompt each one.
How to Use This Knowledge to Write Cold Emails That Actually Get Replies
Start with Claude. Not because it's always the best AI, but because for cold email specifically, its empathy-first default gives you the strongest raw material.
Step 1: Give Claude your brief, but add one sentence about the reader's current situation. Don't just describe who you're targeting — describe what they're feeling right now. For a post-Series A founder: "They're under pressure to show growth, suddenly getting pitched by dozens of vendors, and probably exhausted." That single addition makes Claude's output dramatically more relevant.
A prompt that works: "Write a cold email to a Series A SaaS founder from a freelance conversion copywriter looking to book a 20-minute call. Keep it under 150 words. Avoid all sales language. The founder is currently overwhelmed with vendor pitches and is skeptical of anyone who claims to 'drive growth.' Write like a person, not a pitch deck."
Step 2: Run the output through ChatGPT for structural sharpening. Once you have Claude's empathetic, human draft, paste it into ChatGPT with this prompt: "This is a cold email. Don't rewrite it. Just tighten the structure so the value proposition is clearer in the first two sentences. Keep the tone exactly the same." GPT-4o is excellent at editing — it just shouldn't be doing the first draft for this use case.
Step 3: Test two subject lines using both tools. Subject lines are their own game. Ask ChatGPT for five subject lines under six words. Ask Claude for five that "feel like they were written by someone who already knows the recipient." Compare them. The best one is usually a hybrid — GPT's structural clarity with Claude's human tone.
You can do all three steps in under 20 minutes. The result is a cold email that leads with empathy, lands with clarity, and doesn't sound like it came from a robot — even though it did.
The Part Most People Get Wrong
Most people treat AI cold email tools as a writing shortcut instead of a thinking partner. They paste in a generic brief, take the first output, and wonder why no one replies. That's not an AI problem. That's a brief problem.
The quality of your output is almost entirely determined by the specificity of your input. "Write a cold email to a marketing director" will always produce generic garbage — because you gave the AI nothing to work with. The AI can't know that your prospect just rebranded, is hiring two content managers, or publicly complained about their agency on LinkedIn last week. You have to bring that context.
The other big mistake: using the same AI for every step. People pick one tool and stay loyal to it. But ChatGPT, Claude, and Gemini have genuinely different strengths — and the best cold email workflow uses at least two of them in sequence. Claude for empathy and authenticity. ChatGPT for structure and clarity. Gemini if you need to quickly research your prospect before writing (it handles Google integration and recent web data better than the other two).
Stop asking "which AI is best?" Start asking "which AI is best for this specific step?" That shift alone will double the quality of your output.
Key Takeaways
- Claude's empathy default: Claude produces the most human-sounding cold emails because it naturally centers the reader's emotional state — use it for first drafts.
- ChatGPT's structural strength: GPT-4o is excellent at tightening copy and sharpening CTAs — use it to edit, not to originate.
- Gemini's best role: Gemini 1.5 Pro is better suited for prospect research and summarizing context than for writing persuasive short-form copy.
- Brief quality is everything: A vague brief produces a generic email — add the prospect's current situation and emotional state to unlock dramatically better outputs.
- Multi-tool workflow wins: Using Claude → ChatGPT in sequence produces better results than using any single AI tool from start to finish.
What to Do Right Now
Open Claude right now and paste this exact prompt: "Write a cold email from [your role] to [specific prospect type] who [one sentence about their current situation]. Under 150 words. No buzzwords. Sound like a person." Fill in the brackets with something real from your actual outreach list. Send the result to one prospect today — then try the same brief in ChatGPT and compare what comes back. That 10-minute experiment will teach you more about these tools than any comparison article ever could.