Key Takeaways
- Claude Sonnet 4.6 outperformed Gemini 3.5 Flash at matching the author's personal writing tone in email drafts
- Gemini's questions about context were vague, while Claude asked more relevant follow-ups tied to the actual request
- Claude's speech-to-text engine proved significantly better for users with accents
Claude beats Gemini at writing emails that actually sound like you. That's the verdict from ZDNet's Chandraveer Mathur after a week of testing both AI assistants on real email threads. The difference came down to how each model asked clarifying questions and matched the writer's existing tone.
How the test worked
Mathur compared Gemini 3.5 Flash (free tier, accessed through a Pro account) against Claude Sonnet 4.6 (also free). Both models received identical prompts: draft a follow-up email about shipping delays, request a tracking number, and match the semiformal tone from the existing thread.
To level the playing field, Mathur enabled Google Labs' Personal Intelligence feature, which pulls stylistic cues from prior Gemini conversations. He set Claude's Effort toggle to High. Both received the same two-part instruction: don't respond immediately, and imitate my tone from the thread.
Where Claude pulled ahead
The models diverged immediately in how they sought clarification. Gemini asked two questions. One was useful: what's your target completion date for the review? The second asked whether the recipient had replied since the last email, without explaining why that mattered. The email thread already showed the recipient hadn't provided the tracking number.
Claude asked three questions, each more actionable. Instead of requesting a completion date (which Mathur couldn't commit to without the product in hand), Claude asked for the specific travel date. That's a fixed deadline the recipient can actually work backward from.
More telling: Claude's second question directly asked how firm Mathur wanted to be, explicitly stating the answer would shape the email's tone. Gemini's phrasing was vaguer. Claude also pulled the websites Mathur writes for, suggesting it was reading context more thoroughly.
The speech-to-text gap
Mathur dictates his prompts to save time. Here, Claude's advantage widened. Its speech-to-text engine handled accented English "leagues ahead" of Gemini's, he wrote. For anyone who doesn't type their prompts, this matters.
What this means for enterprise email workflows
Google's Help Me Write ships baked into Gmail, making it the default for two billion users. But convenience isn't the same as quality. If you're sending high-stakes outbound emails, whether sales follow-ups, partnership requests, or media pitches, the AI that best mirrors your voice reduces editing time and avoids the "this sounds like a robot" problem recipients increasingly notice.
The test also highlights a broader pattern. Google optimizes for speed and integration across Workspace. Anthropic optimizes for nuance and instruction-following. Neither is universally better. But for tasks requiring voice preservation, Claude currently wins.
Logicity's Take
This test focused on free tiers, but enterprise buyers should note the pricing delta. Google One AI Premium runs $19.99/month and bundles Gemini across Docs, Sheets, and Gmail. Claude Pro costs $20/month for heavier usage limits but lacks native Workspace integration. For teams already deep in Google's stack, the friction of copying email threads into Claude's interface may outweigh the tone-matching benefits. However, tools like [Zapier](https://logicity.in/r/zapier) or [Make](https://logicity.in/r/make) can bridge that gap by piping email threads into Claude via API, then pushing drafts back into Gmail. Worth exploring if your outbound volume justifies the setup time.
Disclosure
Some links in this post are affiliate links — Logicity earns a commission if you sign up, at no extra cost to you. We only link products we have used or actively recommend.
Should you switch your workflow?
If you're drafting routine internal messages, Gmail's built-in Help Me Write is fast enough. But for customer-facing or high-touch emails where tone preservation matters, running the thread through Claude first produces drafts that need less manual cleanup. The extra step costs time upfront but saves it on the backend.
One caveat: a single user's week-long test isn't a controlled study. Your results may differ depending on your writing style, the complexity of your email threads, and how well you structure your prompts.
Frequently Asked Questions
Is Claude or Gemini better for writing emails?
In this test, Claude Sonnet 4.6 produced emails that better matched the author's existing tone and asked more relevant clarifying questions. Gemini 3.5 Flash was faster but delivered vaguer follow-ups.
Can I use Claude inside Gmail?
Not natively. You'd need to copy email threads into Claude's interface or use automation tools to pipe content via API. Help Me Write is built directly into Gmail.
How much does Claude Pro cost compared to Gemini?
Claude Pro is $20/month. Google One AI Premium, which includes Gemini in Workspace apps, is $19.99/month. Both offer free tiers with usage limits.
Does Claude handle accents better than Gemini for voice input?
According to this test, yes. The author found Claude's speech-to-text engine performed significantly better with accented English.
Another look at how different AI approaches compete on language tasks
Need Help Implementing This?
Setting up AI-assisted email workflows for your team? Logicity's consulting arm helps enterprises integrate LLM tools into existing productivity stacks. Reach out at consulting@logicity.in.
Source: Latest news
Huma Shazia
Senior AI & Tech Writer
Produced with AI assistance and reviewed by the Logicity editorial team. Learn more in our Editorial Policy.
