Gemini 4 Prompt Generator

Write a 1,200-word technical blog post titled 'How Agentic AI Is Changing Software Development in 2026'. The audience is senior software engineers who are sceptical of AI hype. Open with a concrete productivity statistic, then walk through three real workflow changes: automated code review loops, AI-driven debugging, and autonomous dependency resolution. Each section must include a specific before/after example of time saved. End with an honest limitations section covering what agentic AI still cannot do reliably. Tone: authoritative, direct, zero filler. Do not use phrases like 'game-changer' or 'revolutionary'.

Review the following Python function for production readiness. Check for: (1) edge cases and unhandled exceptions, (2) performance issues with large inputs, (3) security vulnerabilities including injection risks and unsafe deserialization, (4) readability and naming conventions, (5) test coverage gaps. For each issue found, provide the exact line number, describe the risk, and write a corrected version of that specific section. Format your response as a numbered issue list, then a full revised function at the end. [PASTE FUNCTION HERE]

Conduct a structured market analysis of the B2B SaaS project management software sector in 2026. Cover: (1) current market size and 3-year growth forecast with sources, (2) the top 5 players by revenue share and their primary differentiators, (3) three underserved customer segments that incumbents are ignoring, (4) the two biggest shifts in buyer behaviour since 2024, (5) a risk matrix covering regulatory, competitive, and technology risks. Present findings in a structured report format with clear headings. Cite sources where available and flag estimates clearly.

Write a 600-word internal product strategy memo for a startup building an AI-powered customer support tool. The memo is from the Head of Product to the founding team. It should recommend whether to pursue a horizontal (serve all industries) or vertical (focus on e-commerce only) go-to-market strategy in the next 12 months. Structure it as: executive summary (3 sentences), market context, recommendation with three supporting arguments, key risks and mitigation, and next steps with owners. Tone: confident, direct, no corporate padding.

Write a Python 3.11 data pipeline that: (1) reads a CSV file from a local path specified in a config dict, (2) validates that required columns ['user_id', 'event_type', 'timestamp', 'value'] are present and throws a descriptive ValueError if not, (3) cleans the data by removing rows where value is null or negative, (4) groups by event_type and calculates count, mean, median, and 95th percentile of value per group, (5) writes the aggregated results to a new CSV and prints a summary to stdout. Use pandas. Include type hints, a docstring for each function, and a main() entry point. No external dependencies beyond pandas and pathlib.

Analyse the following 50-page annual report and extract: (1) the three most significant risks the company disclosed, with the exact page reference and a one-sentence assessment of whether management's proposed mitigation is credible, (2) all forward-looking revenue guidance figures and the assumptions they rest on, (3) any discrepancies between the narrative MD&A section and the financial tables — flag any figure that appears in the text but differs from the audited statements, (4) a sentiment score for the letter to shareholders on a scale of 1–10 (pessimistic to optimistic) with two supporting quotes. Respond in structured sections with clear headers. [PASTE DOCUMENT OR ATTACH FILE]

Build a competitive landscape table comparing five AI writing assistant products: Jasper, Copy.ai, Writer, Notion AI, and Grammarly Business. For each product assess: pricing (entry and enterprise tiers), primary use case, target customer size, key differentiating feature, notable weakness, and estimated number of paying customers if publicly available. After the table, write a 200-word synthesis identifying the clearest gap in the market that none of the five currently addresses well. Use only publicly available information as of 2026 and flag any data that may be outdated.

You are a research agent. Complete the following multi-step task autonomously: Step 1 — Find the five most-cited academic papers on transformer attention mechanisms published between 2022 and 2025. Step 2 — For each paper, extract: title, authors, publication venue, citation count, and the single most important contribution. Step 3 — Identify whether any of the five papers contradict each other on the question of whether sparse attention outperforms dense attention at scale. Step 4 — Based on the evidence, write a 150-word position statement on where the research consensus stands today. Present each step as a clearly labelled section.

Write a 5-email drip campaign for a B2B SaaS company selling automated invoice processing software to finance teams at mid-market companies (100–500 employees). Email 1 (Day 0): problem awareness — open with a specific pain point about manual invoice processing. Email 2 (Day 3): education — explain one key metric the prospect can benchmark internally. Email 3 (Day 7): social proof — a specific customer result formatted as a before/after. Email 4 (Day 12): objection handling — address the most common reason prospects delay purchase ('our current process works fine'). Email 5 (Day 18): low-friction CTA — offer something useful, not just a demo request. Each email: subject line, preview text, 150-word body, and CTA. Tone: peer-to-peer, not salesy.

Implement an LRU (Least Recently Used) cache in Python from scratch — do not use functools.lru_cache or OrderedDict. The implementation must: support get(key) and put(key, value) operations both in O(1) time complexity, accept a capacity parameter at initialisation, evict the least recently used item when capacity is exceeded. Implement using a doubly linked list and a hash map. Include a Node class, the LRUCache class with full docstrings, and a test block in the if __name__ == '__main__' section that demonstrates: cache hits, cache misses, eviction on overflow, and re-insertion of an evicted key. Add inline comments explaining each O(1) design decision.

Write a 15-minute podcast episode script for a show called 'Founders in the Room' — a weekly interview podcast for early-stage startup founders. This episode features a founder who pivoted their B2B SaaS startup three times before finding product-market fit. Structure: 90-second cold open with a surprising statistic about startup pivots, host introduction (20 seconds), guest background (2 minutes), deep-dive on pivot #2 which failed despite good metrics (4 minutes of back-and-forth Q&A), the moment they found PMF (3 minutes), three concrete lessons (3 minutes), sign-off (1 minute). Include specific follow-up questions, not just prompts. Tone: curious, honest, no motivational filler.

Optimise the following SQL query that is running in 8–12 seconds on a PostgreSQL 15 database with 50M rows in the orders table and 2M rows in the customers table. Identify: (1) every missing index that would reduce the query plan cost, specifying the exact CREATE INDEX statement for each, (2) any anti-patterns in the query logic (N+1 patterns, unnecessary subqueries, implicit type casts), (3) whether the JOINs are ordered optimally for the query planner, (4) whether a materialised view would help for this specific access pattern and why. Provide a rewritten version of the query with all optimisations applied and explain the expected performance improvement. [PASTE QUERY AND SCHEMA HERE]

Act as a senior hiring manager at a Series B fintech company. I am preparing for a Principal Product Manager interview in 48 hours. Ask me 8 interview questions in sequence — one at a time — that are specific to fintech product management and reflect what a Series B company actually cares about (not generic PM interview questions). After each of my answers, give me structured feedback: what I did well, what was weak or vague, and a rewritten version of my answer using the STAR format that would score higher. After all 8 questions, give me a 5-point preparation plan for the next 48 hours with specific resources. Start with your first question now.

I am researching the relationship between intermittent fasting and cognitive performance. I have found three studies with conflicting conclusions: Study A (2023, n=120) found 16:8 fasting improved working memory by 15%, Study B (2024, n=85) found no significant cognitive effects, Study C (2025, n=340) found short-term impairment in attention during fasting windows. Synthesise these three findings: (1) identify the methodological differences most likely causing the conflicting results, (2) assess which study design is most externally valid for healthy adults aged 25–45, (3) write a 200-word evidence-based summary a GP could share with a patient asking about fasting and brain health, avoiding both overclaiming and excessive hedging.

Create a brand voice guide for a direct-to-consumer mental wellness app targeting adults aged 28–45 who are functional high-achievers dealing with chronic stress. The guide must include: (1) a one-paragraph brand voice statement, (2) four named voice dimensions with a 2-sentence definition for each and one do/one don't example, (3) a vocabulary section — 10 words/phrases to use and 10 to avoid, with a one-line reason for each, (4) three before/after rewrites of sample UI copy (onboarding prompt, error message, achievement notification) showing the voice in practice. The brand is warm but not saccharine, honest without being clinical, confident without being prescriptive.

Summarise the following SaaS subscription agreement in plain language for a non-lawyer small business owner who is evaluating whether to sign it. Identify and explain in plain English: (1) the auto-renewal clause and the notice period required to cancel, (2) any data ownership provisions — who owns the data the user inputs, and what happens to it after termination, (3) the liability cap — what the maximum payout is if the vendor causes financial damage, (4) the most one-sided clause in the entire agreement and why it is unusual, (5) three questions the business owner should ask before signing. Do not provide legal advice — frame this as 'questions to ask your lawyer'. [PASTE CONTRACT SECTIONS HERE]

Write a 10-minute YouTube video script for a channel called 'Money Mechanics' that explains compound interest to people who find maths intimidating. The target viewer is 22–30 years old with their first real job and no prior finance knowledge. Structure: hook (30 seconds — a surprising number, not a question), the core concept in one analogy that requires zero maths (90 seconds), three real examples with specific dollar figures a viewer on a $55K salary would recognise (4 minutes), the one mistake that wipes out compounding (2 minutes), a 90-second action step the viewer can take this week. No jargon. Every number must be shown on-screen via [B-ROLL: ...] notation. End with a subscriber CTA that is specific and non-generic.

Design a complete n8n workflow (describe each node in sequence) for the following automated process: when a new customer support ticket arrives via Zendesk tagged 'billing issue', the workflow should: (1) extract the customer ID and look up their account status and payment history in Stripe, (2) classify the ticket into one of three categories: payment failure, overcharge dispute, or subscription question — using an AI classifier step, (3) if payment failure: automatically send a templated email from HubSpot with a payment link and pause the ticket for 24 hours, (4) if overcharge dispute: flag for human review, assign to the billing specialist queue, and send an acknowledgement email, (5) if subscription question: generate an AI-drafted reply and send for one-click approval to the support team. Include error handling at each step.

Analyse the attached product photograph for an e-commerce listing audit. Evaluate: (1) technical quality — sharpness, lighting consistency, white balance, and whether the image meets Amazon main image requirements (pure white background, product fills 85%+ of frame), (2) conversion impact — based on visual cues, what questions does this image leave unanswered for a buyer, and what is the single most important missing detail, (3) competitive benchmark — describe how this image would compare against a category-leader listing on visual professionalism, (4) specific retouching recommendations in order of priority. Output as an audit report with a numerical score out of 10 for each of the four areas and a combined score. [ATTACH PRODUCT IMAGE]

Create a 12-week self-study plan to go from beginner to job-ready in data analysis using Python. The learner has: intermediate Excel skills, no prior coding experience, 10 hours per week available, and a goal of landing a junior data analyst role at a tech company. The plan must include: week-by-week topics with specific free resources (name the course or documentation page, not just 'learn Python'), a project milestone every 4 weeks with enough detail to build it, a list of 5 portfolio-ready datasets from public sources, the 12 most important Python and SQL concepts a hiring manager will actually test in an interview, and a realistic time estimate for each major milestone. Flag any assumptions you are making about the learner's pace.