Gemini Prompt Generator — Build Expert Prompts with PromptBro
PromptBro builds structured prompts optimised for Google Gemini — taking advantage of its massive context window, multimodal capabilities, native code execution, and real-time search integration.
What Makes a Great Gemini Prompt?
Gemini 2.5 Pro has one of the largest context windows available — up to 1 million tokens — which makes it uniquely capable for tasks involving extremely long documents, entire codebases, or large datasets in a single prompt. Unlike smaller-context models, you can paste entire reports, transcripts, or source files and ask Gemini to synthesise across them without worrying about truncation. For research tasks, Gemini's integration with Google Search (in the Gemini app) means it can access current information and cite sources — specify "search the web for recent data" when you need up-to-date information rather than relying on the model's training cutoff.
Gemini excels at multimodal tasks — combining text, images, charts, and code in a single prompt. When working with visual data, describe precisely what you want extracted or analysed rather than asking generally. Gemini 2.5 Pro also has native code execution: it can write Python, run it, and return the result in a single turn, making it exceptionally powerful for data analysis tasks. Gemini 2.0 Flash trades some capability for significant speed — use it when you need fast responses at scale. For structured outputs, explicitly request JSON or table format upfront; Gemini follows format instructions reliably when they are stated early in the prompt.
Example Gemini Prompts
Multimodal analysis — image + text
I'm attaching three screenshots of our app's onboarding flow (screens 1, 2, and 3). Analyse each screen and evaluate it against the following UX criteria: 1. Cognitive load: Is the user asked to make too many decisions at once? 2. Progress clarity: Does the user know where they are in the process? 3. Value communication: Is it clear what the user gains from completing this step? 4. Drop-off risk: Are there any friction points that typically cause users to abandon? For each screen, give a score out of 10 for each criterion and a 1–2 sentence explanation. End with a prioritised list of the top 3 improvements across all three screens, ranked by likely impact on completion rate.
Research synthesis — with web search
Search the web for the most current data on AI adoption in the legal industry (2024–2025). I need a research synthesis covering: 1. Which specific legal tasks are seeing the highest AI adoption rates (document review, contract analysis, legal research, drafting)? 2. Which AI tools or platforms law firms are actually using — not just trialling? 3. The primary concerns cited by legal professionals about AI adoption (liability, accuracy, client trust)? 4. Any regulatory or bar association guidance that has emerged in the past 12 months? Format your output as: ## Key Findings (5 bullet points, each with a cited source and date) ## Adoption by Task Type (table: Task | Adoption level | Leading tools) ## Practitioner Concerns (ranked by frequency cited) ## Regulatory Landscape (paragraph) ## What's Still Unclear (gaps in current evidence) Cite all sources with publication name and date. Flag any claims where sources conflict.
Data analysis — with code execution
I'm going to paste a CSV dataset of monthly SaaS revenue figures for the past 24 months. Use code execution to analyse it. Steps to complete: 1. Load the CSV data and display a summary (rows, columns, data types, any nulls). 2. Calculate month-over-month growth rate for each period. 3. Identify the top 3 months by absolute revenue and the top 3 by growth rate. 4. Fit a simple linear trend to the revenue data and project the next 6 months. 5. Plot a chart showing actual revenue, the trend line, and the 6-month projection. After running the code, explain in plain language: - What the growth trend looks like (accelerating, decelerating, stable?) - Whether there are any anomalies worth investigating - How reliable the 6-month projection is given the variance in the data [paste CSV data below]
Long document summarisation — 1M context
I'm attaching the full text of a 400-page annual report. Use your full context window to read the entire document before summarising — do not truncate your reading. Produce the following outputs: 1. Executive Summary (250 words): Key financial performance, strategic direction, major risks. 2. Financial Highlights Table: - Revenue (this year vs. last year, % change) - Operating profit / EBITDA - Free cash flow - Key segment breakdown 3. Strategic Priorities: What are the 3–5 things leadership is explicitly betting on for the next 3 years? 4. Risk Register: List every material risk the company discloses, with their stated mitigation. 5. Things That Seem Off: Flag any discrepancies, unusually hedged language, or figures that don't add up across sections. Important: base every claim on specific passages from the report. Quote the relevant section and page number where possible.
Build your own Gemini prompt with PromptBro
PromptBro's 6-step guided flow builds prompts optimised for Gemini's specific strengths — massive context windows, multimodal reasoning, code execution, and real-time search. Describe your goal and PromptBro structures it for maximum Gemini performance.
Try PromptBro free — build your first prompt in 60 seconds →