Think Hard's prompt condenses a lengthy support conversation for a colleague, such as an engineer or a manager, who has seen none of it. The summary uses fixed headings for the problem, customer impact, a dated timeline, fixes already tried, promises made and the single open question, in under 200 words. Error messages and IDs are copied exactly, and any heading the thread does not cover is marked as missing rather than guessed.
Summarize the support thread below for {{escalation_team}}, who have not seen any of it. They should be able to act on your summary without reading the thread.
Thread:
"""
{{thread}}
"""
Use this format, with these exact labels:
Problem: one sentence on what is broken or being asked, in plain words.
Customer impact: who is affected and how badly, using only facts from the thread.
Timeline: up to 5 dated bullets of what happened, oldest first.
Already tried: each fix that was attempted, and what happened.
Promised to the customer: any commitment made, by whom and when.
Open question: the one decision or answer needed from {{escalation_team}}.
Keep it under 200 words. Copy error messages, order numbers and IDs exactly as written.
Leave out greetings, apologies and back-and-forth that does not change the facts, and do not speculate about the cause.
The summary should be usable after one read: for example, an engineer should be able to start reproducing a bug straight away.
If the thread is unclear or silent on one of the labels, write "Not stated in the thread" for it. Do not guess.
Calm an upset customer with a direct, fair replyThis prompt drafts a support reply to an upset customer from the facts you supply: what actually happened, what you can offer and what you cannot do. It opens by naming the customer's problem in their words, admits fault plainly when you were at fault, gives a concrete fix and date, and ends with one next step. It bans stock phrases, keeps chat replies under 150 words and email under 200, and also points out the sentence most likely to make the customer angrier.Customer support
Turn a support ticket into a known-issue articleThis prompt asks the AI to write a short article for a support knowledge base from a closed ticket conversation. The article has two parts, a short summary of the problem and the fix that was actually provided, with no names of people or companies. It tells the AI to double-check the result for any names before finishing.Customer support
Write a 200-word brief for a booked sales callPrompt A from Pipe Playbook turns the history you already have on a booked call into a brief of under 200 words. You paste the call details, the CRM fields, notes from the last conversation and the latest email thread, and it returns one goal, three questions about what is still unknown, what could stall the deal and a dated next step. It marks gaps as unknown and labels guesses as inferences so you can check or delete them.Sales
Build a website that does not look AI-madeThis prompt asks an AI coding or design assistant to build a site or page while avoiding the visual habits that give away a site generated by AI. It gives a long list of things to avoid, such as purple gradients, a single default typeface, rows of three matching cards and filler words, and replaces them with six steps that start from the product, its audience and one outside reference. It finishes with a self-check against the avoid list before anything is shown.Coding agent
Build evidence-backed customer personas from reviews and surveysThink Hard's system prompt and user message turn customer data such as reviews or survey answers into two to four personas for a named team and decision. Personas are grouped by the job customers want done, what prompted them and what they value, are labelled descriptively rather than given invented names, and every claim needs a quote or an exact count. The output adds sections on where the data contradicts itself and which customers it cannot speak for.Marketing
