India-based. Globally connected.   Online · On-site · Hybrid

AI Prompt Library

A better starting point
for everyday work.

Replace the bracketed fields with permitted context. Never include sensitive data in an unapproved tool; always review the result.

Find a prompt for your task

Sales

Draft relevant outreach

Using only the product facts and fictional prospect context below, draft a concise first-contact email. Explain one relevant business use. Do not invent personalization, outcomes or customer references. Mark missing facts as [NEEDS INPUT]. Product facts: [insert approved facts] Prospect context: [insert non-sensitive context]

Marketing

Turn a brief into content

Create three campaign concepts for [audience] using the approved brief below. For each, include the message, channel and supporting fact. Follow [brand voice]. Do not add statistics or unverified claims. List the claims a human reviewer should check. Brief: [insert]

HR

Explain a policy clearly

Rewrite the approved policy below as a clear employee announcement. Preserve obligations, exceptions and dates exactly. Do not infer policy. Identify ambiguous passages for HR review. Include a short FAQ based only on the supplied text. Policy: [insert approved text without personal data]

Finance

Review a reporting narrative

Using the synthetic figures below, draft a variance narrative. Show the arithmetic behind each comparison and separate facts from possible explanations. Do not invent causes. Flag missing data and recommend verification steps. Figures: [insert synthetic data]

Management

Prepare a decision brief

Turn the supplied notes into a one-page decision brief with context, options, evidence, risks, open questions and a proposed next step. Attribute claims to the notes. Keep unknowns explicit and do not invent commitments. Notes: [insert approved notes]

Operations

Draft a process checklist

Convert the approved process description into a checklist with inputs, steps, owner, review point and escalation path. Flag missing controls rather than inventing them. Label the output DRAFT FOR PROCESS OWNER REVIEW. Process: [insert]

Engineering

Find gaps in requirements

Review the requirements below for ambiguous terms, missing acceptance criteria and untested edge cases. Produce a table with issue, source passage, question and proposed test. Do not invent technical constraints. Requirements: [insert permitted material]

Research

Build an evidence plan

Create a research plan for [question]. Suggest primary-source types, search queries and criteria for comparing evidence. Keep dated facts separate from opinion. Never fabricate citations. If you can access sources, provide links and state what each supports.

Emails

Make an email clearer

Rewrite the draft below for [audience] with a clear purpose, concise context and one explicit next action. Preserve names, facts and commitments. Mark unclear details [CONFIRM]. Offer a neutral subject line. Draft: [insert non-sensitive draft]

Meetings

Extract decisions and actions

From the permitted meeting notes below, extract decisions, actions, owners, deadlines and unresolved questions. Use a table. If an owner or deadline is absent, write Not specified. Do not infer agreement from discussion. Notes: [insert]

Excel

Explain and test a formula

Help me build a spreadsheet formula for [task] using the sample columns below. State assumptions, explain each part, and provide normal, blank and error test cases with expected results. Use synthetic data and do not claim the formula is verified. Sample columns: [insert]

Presentations

Create a decision-led outline

Draft a [number]-slide outline for [audience] to decide [decision]. Use only the approved brief. For each slide include a headline, key point and needed evidence. Mark evidence gaps [CONTENT REQUIRED]. Do not invent metrics. Brief: [insert]

From prompt to reviewed workflow

How do you adapt a prompt to your organization?

Replace each bracketed field with the information needed for the task. State the intended audience, approved source material, required output and what the model should do when information is missing. Do not paste confidential details into an unapproved tool simply because a prompt asks for context.

Test the prompt on a small permitted example and review the result before saving it for reuse. A prompt that works once may fail when the input is incomplete or ambiguous. Add the review checklist to the workflow so the team can judge quality consistently.

Learn prompt engineering