Gemini 3.5 Pro Prompt Generator

Use Deep Think mode. Synthesise the current state of research on CRISPR-Cas9 off-target editing in human somatic cell therapy as of 2025–2026. Your output must cover: (1) The three most cited categories of off-target mechanisms, with a brief explanation of each; (2) The two detection methods with the best sensitivity-specificity tradeoff — name the assay, describe its principle, and cite the key validation study; (3) Where scientific consensus currently stands on whether off-target rates are clinically acceptable for ex vivo therapies vs. in vivo delivery; (4) The two most prominent open scientific disputes — describe each disagreement, the leading camps, and what evidence is needed to resolve it; (5) A 150-word plain-language summary suitable for a hospital ethics committee with no molecular biology background. Flag any claim where evidence quality is weak or primarily from preprints.

I am attaching a full corporate legal disclosure package (200+ pages: 10-K, proxy statement, SEC comment letters, and earnings call transcripts for the last 4 quarters). Analyse the complete package and extract: (1) Every forward-looking statement made across all documents — table format: Document | Quarter | Statement | Outcome if verifiable in a subsequent document; (2) Any material discrepancy between management's verbal guidance on earnings calls and the written disclosures in the 10-K; (3) The evolution of risk factor language across all 4 quarterly filings — identify any new risks added, risks removed, or risks reworded in ways that soften their severity; (4) Compensation committee decisions flagged in the proxy that are unusual relative to peer group disclosures; (5) A 250-word executive summary of the most significant legal or financial exposure identified across the full package. [ATTACH ALL DOCUMENTS]

Analyse the following enterprise SaaS Master Service Agreement from the perspective of a general counsel at a mid-market company (500 employees, B2B SaaS product, $50M ARR). Identify and explain in plain English: (1) Every provision that deviates from market-standard terms in a way that favours the vendor — flag severity as Critical, Major, or Minor; (2) The indemnification clause: who indemnifies whom, for what, and is the carve-out for IP infringement standard or one-sided?; (3) The data processing and security provisions: do they meet the minimum threshold for a GDPR data processing agreement, and is the sub-processor disclosure mechanism workable?; (4) Termination for convenience rights: does the customer have them, and at what notice period?; (5) Auto-renewal and fee increase provisions: what triggers them, and is the notice window typical?; (6) The three redlines you would propose as highest priority before signing. Do not give legal advice — frame as 'questions to raise with outside counsel' where appropriate. [PASTE CONTRACT]

Build a 3-year financial model for a B2B SaaS startup with the following base-case assumptions: $2M ARR at the start of Year 1, 80% net revenue retention, average contract value $24,000/year, 18-month average sales cycle, 3 enterprise AEs each closing 8 deals/year, $600K annual infrastructure cost scaling at 30% per year, headcount at 40 with $3.2M total comp growing 25%/year. Model three scenarios — (A) Base case as stated, (B) Bull case: NRR improves to 115%, AE team grows to 5 in Year 2, ACV increases 20% through upsell, (C) Bear case: 6-month sales cycle slowdown in Year 1 reduces new bookings 40%, NRR drops to 72%, one AE departs in Q2 Year 2. For each scenario, output: ARR by year-end, gross margin, burn rate by quarter, runway assuming a $5M Series A raised at start of Year 1, and the key operational metric that differs most between scenarios. Clearly label assumptions. Format as a structured financial report with tables.

Design a rigorous mixed-methods research study to answer the following question: 'Does AI-assisted code review improve software quality outcomes (as measured by production defect rates and mean time to resolution) at enterprise software teams of 50+ engineers?' The study design must include: (1) Research question refinement — identify 3 operationalisable sub-questions and the primary outcome variable; (2) Study design rationale — why mixed methods rather than purely quantitative or qualitative, and which specific design type (explanatory sequential, exploratory sequential, concurrent triangulation) is most appropriate and why; (3) Quantitative component — sample size calculation with power analysis at 80% power (state your assumed effect size and why), randomisation strategy, control group design, data collection instruments, and primary statistical test; (4) Qualitative component — participant selection criteria, data collection method, and analysis approach (thematic analysis, grounded theory, etc.) with rationale; (5) Threats to internal and external validity and how each is mitigated; (6) Ethical considerations specific to workplace research involving proprietary code. Format as a methodology chapter draft.

Analyse the competitive positioning and durable moat of Salesforce (CRM) as of 2026, using Hamilton Helmer's 7 Powers framework. For each of the 7 Powers — Scale Economies, Network Effects, Counter-Positioning, Switching Costs, Branding, Cornered Resource, Process Power — assess: (1) Whether Salesforce currently possesses this power (Strong / Partial / Absent), with a 2–3 sentence justification citing specific product, pricing, or market behaviour; (2) The primary threat to this power in a 3-year horizon (name the specific competitor or technology); (3) A single metric or observable that would indicate this power is strengthening or weakening. After the 7-power analysis, write a 200-word synthesis: given the full picture, what is the most likely scenario in which Salesforce's market leadership is meaningfully eroded by 2029, and what early warning signal would a strategic analyst watch for?

Write a regulatory impact assessment for a hypothetical EU AI Act Article 50 compliance programme at a mid-sized European fintech (1,200 employees, offers credit scoring and fraud detection AI to banks). The assessment must cover: (1) Classification analysis — which of the company's AI systems fall under which risk category (Unacceptable, High, Limited, Minimal) under the EU AI Act, with a one-paragraph justification for each classification; (2) For each High-risk system: the specific obligations under Articles 9–15 (risk management, data governance, transparency, human oversight, accuracy and robustness) and a gap assessment against assumed current practices of a typical mid-market company; (3) Compliance programme roadmap — phased over 18 months, with Q1–Q2 foundational actions, Q3–Q4 technical implementation, and Year 2 ongoing monitoring; (4) Estimated compliance cost range with key cost drivers; (5) Residual risks after full compliance and the two regulatory ambiguities that require legal interpretation. Format as an internal board memo.

Review the following distributed system architecture description for correctness, scalability, and failure mode resilience. After the review, provide: (1) An assessment of the CAP theorem tradeoff this architecture implicitly makes — is it CP or AP, and is that the right choice for the stated use case?; (2) Every identified single point of failure — component name, failure scenario, blast radius (which services are affected), and the mitigation pattern that should be applied (circuit breaker, retry with backoff, async queue decoupling, etc.); (3) The two most likely latency bottlenecks under a 10× traffic spike — identify the bottleneck, explain why it fails at scale, and propose the architectural change; (4) Data consistency risks — any point in the system where a network partition or process crash could leave data in an inconsistent state, and the recovery strategy needed; (5) A prioritised list of 5 architecture changes, ordered by impact-to-effort ratio. [DESCRIBE YOUR ARCHITECTURE]

Write a 600-word policy brief on the proposed UK Digital Markets, Competition and Consumers Act provisions on AI foundation model oversight. The brief is for a parliamentary committee reviewing the legislation. Structure: (1) Executive summary (50 words) — the core policy question and the brief's position; (2) Background — what the provisions require, which companies they target, and how they differ from the EU approach; (3) Stakeholder analysis — table format: Stakeholder | Position | Key Argument | Vested Interest | Credibility Score (High/Medium/Low based on evidence quality); (4) Analysis of the three most contested provisions — for each, present the strongest argument for and the strongest argument against, then give a one-sentence assessment of which is more supported by evidence; (5) Recommendation with two alternative policy options and their tradeoffs; (6) Implementation risk matrix — 4 risks, likelihood, impact, and mitigation. Cite real legislation where relevant. Flag any claims that are contested.

Build a production-ready async distributed task queue in Python 3.12 using asyncio, Redis (via aioredis), and Celery as a reference architecture, but implemented from scratch without Celery. Requirements: (1) A Task class that stores task function, args, kwargs, priority (1–5), retry count, and max retries; (2) A Queue class backed by Redis sorted sets — priority as the score, task ID as the value; (3) An async Worker class that: polls the queue with configurable concurrency limit (default 4 workers), executes tasks using asyncio.create_task, handles task success (remove from queue, log to Redis hash), handles task failure (increment retry count; re-enqueue if under max_retries, move to dead-letter queue if exceeded), implements exponential backoff between retries (base 2, max 60 seconds); (4) A simple decorator @task(priority=3, max_retries=3) that registers functions as queue-aware tasks; (5) Graceful shutdown on SIGTERM — drain in-flight tasks, stop accepting new ones. Provide complete, runnable code with type hints throughout and a minimal example in __main__.

Write a structured investment thesis for a 3-year position in a hypothetical company: a vertical SaaS provider for independent veterinary clinics (450 active customers, $12M ARR, 115% NRR, $2.4B TAM, 2 well-funded competitors). Structure the thesis with three fully developed scenario cases: (1) Bull case — what must be true about market expansion, competitive differentiation, and operational leverage for the company to 3× ARR in 3 years; what multiple does this justify and why; what is the most credible path to each assumption; (2) Base case — 2× ARR, identify the one or two constraints that prevent the bull case, and the multiple this scenario justifies; (3) Bear case — stagnation or ARR decline; which competitor move, customer churn dynamic, or macro event causes it; what is the floor valuation and why; (4) Key investment risks ranked by probability × impact; (5) The three KPIs you would monitor quarterly to determine which scenario is unfolding, with the specific threshold that would cause you to exit the position. Format as a VC investment memo.

Write a 900-word academic research grant proposal for submission to the UK Research and Innovation (UKRI) Future Leaders Fellowship programme. Research topic: 'The cognitive and neural mechanisms underlying human trust calibration in AI-assisted decision-making across high-stakes domains (medicine, law, finance).' The proposal must include: (1) Scientific significance — why this question matters, the current knowledge gap, and what prior work the applicant builds on (cite 3 plausible but fictional prior studies with realistic author names, journals, and years); (2) Research objectives — 3 specific, testable objectives aligned with the 4-year fellowship timeline; (3) Methodology — one paragraph per objective: design, participants, instruments, analysis plan; (4) Expected outcomes and their significance — what new knowledge is produced and how it advances the field; (5) Broader impact — who benefits beyond academia and how findings will be translated; (6) Feasibility — why this PI is uniquely positioned to do this work. Tone: formal, active voice, confident without overclaiming. Mark any claim that would need real citations as [CITE].

Build a comprehensive due diligence checklist for a proposed strategic acquisition of a Series C AI infrastructure startup ($45M ARR, 120 employees, primary product: GPU orchestration and job scheduling software for ML training workloads). The acquiring company is a Tier 1 cloud provider. Organise the checklist into 8 workstreams: (1) Technology & IP — source code review, patent landscape, open-source licensing obligations, technical debt assessment criteria; (2) Commercial — customer concentration, contract terms, NRR by cohort, pipeline quality; (3) People & Culture — key person dependencies (who leaves = significant product risk), retention risk, equity waterfall; (4) Financial — revenue recognition policy, deferred revenue, any non-recurring revenue, working capital requirements; (5) Legal & Regulatory — IP ownership chain, pending litigation, export control compliance (ITAR/EAR for GPU tech), data residency obligations; (6) Security & Compliance — SOC 2 status, penetration test history, incident response record; (7) Integration complexity — APIs, data models, customer migration path; (8) Antitrust risk — market share implications, any prior regulatory interaction. For each workstream, include 5–8 specific questions with the document or data room artefact that would answer each.

Design and implement a complete dimensional data model for an e-commerce analytics platform that must support the following business questions: (1) What is the 30/60/90-day customer lifetime value by acquisition channel and first-purchase category?; (2) Which product combinations are most frequently bought together within the same order, and has this changed quarter-over-quarter?; (3) What is the median time from cart creation to purchase, segmented by device type, customer tier, and applied discount?; (4) Which supplier's products have the highest return rate, and is the return reason correlated with shipping carrier? Deliverables: (1) A fully specified star schema — fact table(s) and dimension tables with column names, data types, primary and foreign keys, and a one-line description of each column's business meaning; (2) The ETL logic for the most complex slowly-changing dimension in the model (SCD Type 2); (3) SQL queries answering each of the 4 business questions, optimised for a Snowflake environment; (4) An explanation of any design tradeoff you made between query performance and storage efficiency.

I am attaching a 120-page management consulting report on digital transformation in the UK National Health Service. The report covers 14 recommendation areas across technology, workforce, procurement, and governance. Transform this document into: (1) A 300-word executive summary for the Secretary of State for Health — one paragraph per the three most actionable recommendations, in plain English with zero NHS acronyms unexplained; (2) A one-page briefing note for an NHS trust CEO — what it means for their organisation specifically, what they are expected to do in the next 12 months, and what funding or policy changes they should watch for; (3) A 10-slide presentation outline — slide title, one-sentence key message per slide, and what visual (chart, table, diagram) would best communicate each message; (4) A press release (250 words) for the Department of Health — lead with the most newsworthy finding, include one attributed quote (write the quote, mark as [REVIEW]), and end with context about the scale of the transformation programme. [ATTACH REPORT]

Conduct a structured threat modelling exercise using the STRIDE methodology for the following web application architecture: a multi-tenant SaaS application with a React frontend, Node.js API layer, PostgreSQL database, Redis session store, and S3 for file storage. All components are deployed on AWS. Identify for each STRIDE category — Spoofing, Tampering, Repudiation, Information Disclosure, Denial of Service, Elevation of Privilege: (1) The specific attack vectors applicable to this architecture, with the exact component and data flow targeted; (2) The mitigating control that should be in place (name the specific AWS service, security header, library, or configuration); (3) How to verify the control is correctly implemented (specific test or audit step); (4) The CVSS v3.1 base score range you would assign to the unmitigated risk and why. After the STRIDE analysis, identify the three highest-priority remediation actions for a team starting a security programme with limited budget, and explain the reasoning for that prioritisation.

Design a structured prior art search framework for a patent application covering a novel method for real-time on-device AI model compression during inference on mobile hardware. The invention claims: (1) A dynamic pruning algorithm that adjusts sparsity thresholds based on available memory headroom, (2) A hardware-aware quantisation scheduler that profiles thermal throttling events to pre-emptively reduce model precision, (3) A layer-skipping controller triggered by confidence scores in early exit architectures. Provide: (1) Search strategy — the 8 most important IPC/CPC classification codes to search, with rationale for each; (2) Key technical terms and synonyms to use in each claim element's keyword search — formatted as a Boolean search string per claim; (3) The 5 most likely prior art databases (patent and non-patent literature) and why each is relevant to this technology area; (4) A structured evaluation rubric for assessing whether a found reference anticipates or merely renders obvious each claim; (5) The three questions a patent attorney would want answered before concluding the search is sufficiently exhaustive.

Conduct a structured scenario planning exercise for a mid-market newspaper group (8 regional titles, 2M monthly readers, 60% digital revenue, 40% print) planning its 5-year strategy in 2026. Follow the Shell scenario planning methodology: (1) Identify the two most critical and most uncertain driving forces shaping the industry over 5 years — justify why these two (not others) are the pivotal uncertainties; (2) Construct four named scenarios by crossing the two axes — for each scenario: give it a vivid name, describe the world in 2031 in 2 paragraphs (what happened in media, advertising, politics, and technology), and describe what the newspaper group's situation looks like in that world; (3) For each scenario, identify the three strategic options that would be most effective and the three options that would destroy value; (4) Identify the two or three strategic moves that are robust across all four scenarios — these become the core strategic commitments; (5) Define 5 early indicators (observable within 12 months) that would signal which scenario is emerging. Format as a board strategy document.

I am providing a dataset of planning application decisions from a UK local authority over 10 years (as a CSV). Analyse the data and construct the outline for a data-driven investigative news article. Provide: (1) The three strongest data-supported angles — for each, describe the finding in one sentence, the supporting statistics, and the public interest case; (2) For the strongest angle: a full article outline — headline, standfirst (3-sentence summary), introduction paragraph, 5 body sections with main argument and key supporting data point per section, expert sources to seek (describe the type of expert, not fabricated names), and conclusion; (3) A statistical validity check — what are the three most important caveats a data editor would flag about the analysis (confounders, data quality, selection bias)? (4) The two Freedom of Information requests that would significantly strengthen the story if answered; (5) A list of 4 right-of-reply questions for the local authority press office. [ATTACH CSV DATA]

Apply RICE scoring (Reach, Impact, Confidence, Effort) to prioritise the following 8 product initiatives for a B2B analytics platform with 3,000 active users and a 10-person engineering team. For each initiative, make explicit assumptions for each RICE factor and show your calculation. Initiatives: (1) Real-time dashboard auto-refresh, (2) CSV export for all report types, (3) Custom calculated metrics builder, (4) Slack/Teams alert integration, (5) White-label embedding for enterprise clients, (6) Mobile app (iOS only), (7) AI-generated insight summaries on every dashboard, (8) SSO/SAML integration. After scoring all 8: (1) Present the full RICE ranking table; (2) Identify any initiative where the score is particularly sensitive to your confidence estimate — where a higher-confidence validation would significantly change the ranking; (3) Flag any initiative that the RICE score underweights for strategic reasons (e.g. an enterprise-contract prerequisite); (4) Recommend the optimal first-quarter scope for a 10-person team and explain the tradeoffs. Show all RICE calculations explicitly.