Enhancing Your Experience with Google Gemini
In today’s digital landscape, many of us seamlessly integrate Google Gemini into our daily routines, whether we’re composing emails in Gmail, drafting documents in Google Docs, or browsing the web. Yet, the prompts we provide to this AI assistant can significantly influence the outcomes we receive. A vague prompt often yields generic responses, while a more detailed instruction allows Gemini to generate more tailored results.
The encouraging news is that Gemini thrives on structure. By clearly defining the task, offering relevant context, and specifying the desired format for the response, you can enhance the utility of its output. Below are some strategies for crafting more effective prompts for Google Gemini.
Utilising the Four-Part Formula
Google suggests structuring requests around four key elements: persona, task, context, and format. Begin by stating who you are or what role you want Gemini to adopt, then outline the task at hand, provide pertinent background information, and clarify how you would like the response to be presented. While you may not need to include all four elements every time, incorporating them can lead to more relevant and useful replies.
For instance, instead of saying: “Draft an email announcing our new flexible work policy,” consider phrasing it as: “You are the Human Resources manager at a tech company with 50 employees. Compose an email to announce that employees will be allowed to work remotely two days a week. Maintain a friendly yet professional tone, explain that supervisors will approve schedules, and include three key points summarising the policy. Limit the email to 200 words.” In this example, Gemini assumes the role of HR manager, with a clear task, context, and specified format.
Incorporating Tags for Clarity
For more complex prompts, Google recommends using headers to differentiate instructions from supporting material. This can be done with Markdown headers like #Header 1 or XML-style tags such as and . Such tagging helps Gemini discern between the prompt’s instructions and reference material, maintaining the organisation of longer prompts.
For example, instead of pasting a lengthy article followed by “Summarise this,” you might say:
#Task Summarise the article below into five key points for a general audience. Retain all figures and note any claims that lack proper attribution.
Then, insert the article text. This structure clarifies the task and sets expectations for the response.
Creating a Gem for Reusability
A ‘Gem’ allows you to store instructions and reference materials, creating a reusable version of Gemini. This way, you won’t need to replicate the same context each time you initiate a new chat. You can configure it with your preferred style, recurring tasks, or specific response formats, making it easy to access when needed.
Establishing Basic Rules in Settings
The “Prompts for Gemini” feature lets you specify how you want the assistant to respond during chats. You can set preferences for tone, vocabulary, format, and other recurring requirements, which you can later edit, disable, or remove as necessary. While this doesn’t guarantee improved responses, it saves you from having to repeat the same instructions in every conversation.
Utilising Temporary Chat Features
Gemini’s version of incognito mode is known as “Temporary Chat.” These conversations do not appear in your recent chats or in the “Gemini app activity,” and Gemini does not use them for personalising future interactions. Features that rely on personalisation, including Gems, are unavailable here, meaning responses may be less tailored. However, “temporary” does not imply immediate deletion, as Google indicates that chats can be retained for up to 72 hours for service provision and feedback processing.
Engaging in Depth Research
The “In-Depth Research” feature is tailored for more complex queries rather than quick searches. Once you submit your research question, Gemini will generate a proposal for you to review and edit before proceeding. Use this phase to clarify what you would like it to investigate, which sources to prioritise, and how you wish the final report to be structured.
For example, rather than simply stating: “Research electric vehicles in Mexico,” try: “Investigate the adoption of electric vehicles in Mexico and Latin America since 2023. Compare sales, charging infrastructure, government incentives, and major barriers to adoption. Prioritise government data, information from car manufacturers, and peer-reviewed studies over promotional material. Distinguish confirmed figures from estimates and present the findings in a 1,000-word report with a comparative table and linked sources.” Before commencing the research, review Gemini’s proposed plan and make adjustments to include any overlooked markets, source categories, or comparison points.
Maximising the Use of Canvas
Canvas provides an independent workspace for drafting and reviewing documents, code, and other projects. You can edit materials directly, highlight specific sections for Gemini to modify, or adjust elements like length and tone without starting from scratch.
Speaking Instead of Writing
If you find it challenging to condense a complex idea into a clear written prompt, consider articulating it aloud. Speaking can help you convey context, clarify your objectives, and include details you might otherwise overlook. Just remember, a longer voice prompt does not automatically mean better quality; the task and constraints still need to be clear.
Streamlining Your Workflow
When you have applications connected to Gemini, it can leverage information from services like Gmail, Drive, and Calendar. Additionally, you can use the “@” menu to directly select an application or source, rather than instructing it where to look. Review these connections in the “Settings” menu and deactivate any that you do not wish Gemini to access.