In a recent test comparing the email-writing capabilities of two large language models—Gemini by Google and Claude by Anthropic—both tools were found to be effective in drafting contextually appropriate emails. Gemini is the engine behind Google's "Help me write" feature, which has been integrated into Gmail and Google Docs for three years. This feature allows users to draft entire emails with minimal input, requiring only validation from the user. However, the user conducting the test noted that while "Help me write" is a powerful tool, the underlying large language model (LLM) performs better when used through Google's autonomous AI applications. To ensure a fair comparison, the user tested Gemini 3.5 Flash, a faster model accessible via a Pro account, and activated Google's "Personal Intelligence" option in Google Labs. For Claude, the user used the free Claude Sonnet 4.6 model, setting the "Effort" parameter to "High" for optimal performance. The user also took steps to streamline the process, using clear instructions to avoid unnecessary interactions and to focus on the email's objective. These instructions included asking the AI not to respond immediately, to avoid follow-up questions, and to mimic the user's tone based on the email thread. The test involved an email about a shipping delay and the lack of a tracking number for a product. The prompt asked the AI to draft a response requesting the tracking number again and to explain the need for a deadline for product evaluation before departure. Gemini's draft included two questions, one about the expected completion date and another about whether the recipient had responded to a previous email. However, the second question was redundant, as the email thread already mentioned a previous request for the tracking number. In contrast, Claude's draft asked three questions, including one about the desired level of firmness in the tone, which directly addressed the email's tone. The third question from Claude extracted websites from the user's email signature and asked where the review would be published, a detail already mentioned in a prior conversation. The two responses were notably different in tone and structure. Gemini's draft covered all essential points but was verbose and used a more formal tone, which felt less urgent for the second request. Its length slightly diluted the sense of urgency. Claude's response, on the other hand, was concise and perfectly matched the user's preferred tone. It included personalized touches and relevant follow-up questions, creating a more effective draft. Although the "Send via Gmail" button appeared in Claude's response due to the user logging in with a Gmail ID, this feature is typically more associated with Gemini, which is more deeply integrated into Google's ecosystem. The user concluded that while Gemini benefits from these integrations, it fell short in some areas, especially in directly addressing the prompt. Claude, with its simplified setup through activated integrations in the "Skills" section, was ultimately preferred for its precision and ease of use.