A party rental firm cut reply time from four hours to under 10 seconds, and bookings rose 35%.
The prompt · from the source
You are a lead qualification assistant for a local service business.
Lead info:
Name: {{ name }}
Message: {{ message }}
Source: {{ source }}
Respond with ONLY a JSON object:
{
"score": <1-10>,
"intent": "<booking|question|price_check|spam>",
"urgency": "<high|medium|low>",
"suggested_reply": "<personalized 2-3 sentence reply>",
"follow_up_delay_hours": <24|48|72>
}
Score guide: 8-10 = ready to buy, 5-7 = interested, 1-4 = cold/spam.
A freelancer got tired of chasing late payers, so he built three small agents to do the awkward part for him.
Try it · based on the source's workflow
Chase overdue invoices for [business]. First, scan [invoice list] and rank each overdue invoice by amount and days late. Draft a reminder for each one: polite at [3] days, firm at [30] days, formal at [60] days, matched to [email, text, or WhatsApp]. Read each reply. If a client promises payment, pause reminders for that client for [48 hours]. Send nothing yet. Show me every draft for approval first, because one wrong detail can make a good client angry.
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
An AI caller reached each new lead in under 10 seconds and booked 215 appointments in two weeks.
Try it · based on the source's workflow
When a new lead arrives from [lead form], call them within [10 seconds]. Use the form data: name, [fields]. Ask if they want to book an appointment, and offer times from [calendar]. If the call goes to voicemail, follow up [number] times at [intervals]. Book in [calendar tool] and log the result in [system]. Show me a daily list of calls, bookings, and leads you could not reach.
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
An AI audited 912 patient charts, and 45 patients booked treatment worth $27,205 in production.
Try it · based on the source's workflow
Audit the client records in [practice system]. For each chart, check the treatment plan, the clinical notes, and the ledger. List only the clients with care that is still open and still valid, with the treatment, the estimated cost, and the last visit date. Rank the list by value and by time since last visit. Give the list to [coordinator] for outreach. Do not contact clients yourself.
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
A WhatsApp agent answers from live listings, checks budget first, and books the inspection itself.
The prompt · from the source
You are a helpful Real Estate Assistant. Your goal is to get the user to book an inspection. Before booking, you must politely ask for their budget and timeline. If they cannot afford the property, politely suggest cheaper alternatives.
One outbound system booked 38 meetings in seven days, and its best message variant replied at 23%.
Try it · based on the source's workflow
Run outbound for [business]. Build a list of [number] likely buyers from [sources] and sort them into plays. For each account, find one real signal in [notes] and draft three message variants in my voice. Keep one record per person across [channels] so a reply on one stops the other. Sort replies into needs me, engaged, and meeting booked. Show me every draft before anything sends.
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
Three linked workflows file her email and invoices alone, and missed follow-ups fell to zero.
Try it · based on the source's workflow
Run my email admin in [automation tool]. For each new email in [account]: save attachments to [drive folder], create a record in [database] linked to the right client in [CRM], and label the message as done. Sort each email into one of [categories] and pull out tasks with due dates into [task list]. For each invoice PDF, read the amount, date, and sender, and file it in the right [book or folder]. Show me anything you cannot file, and send me a daily summary.
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
A solo dome maker scored 1,051 CRM leads for $15.07 and found $82,998 in pipeline.
Try it · based on the source's workflow
Score the leads in [CRM]. For each contact, read the full email history, check the linked company, and visit their website. Score each lead as hot, warm, or cold for [business], using these signals: [hot signals] and [cold signals]. Write the score and a one-line reason back to [CRM]. Give me a call list in order, with what to say to each lead. Do not contact anyone.
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
A solo SaaS owner's support agent cut reply time from four hours to eight minutes in eight weeks.
Try it · based on the source's workflow
Act as the support agent for [business]. Read each new message in [support inbox] and sort it: [login, billing, feature request, bug, account change, refund, general, escalate]. Pull the facts first from [payment system, account records, issue tracker]. Reply in my tone from these examples: [paste]. You may [send reset links, refund under amount, update billing]. Never delete an account. Draft and flag anything angry, complex, or outside these rules for my review. Send me a daily digest.
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
A four-person plumbing firm cut missed after-hours calls from 11 to 14 a week to two or three.
Try it · based on the source's workflow
Answer calls for [business] when we cannot: after [time], on [days], and after [four] rings. Use only our service area [area], our rates [rates], and these call classes: [classes]. For an emergency ([your emergency cases]), text [contacts] at once with the caller's name, address, and problem. For other jobs, book the next open slot in [scheduling tool]. Never invent a slot or a price. Take a full message for anything you cannot answer.
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
Job postings became the buying signal, and 311 LinkedIn connections turned into 26 meetings.
Try it · based on the source's workflow
Find leads for [business] from job postings. Watch [job boards] for firms hiring for [roles my product helps with]. For each firm, write a short LinkedIn message for the [executive, decision maker, champion] that names the posting and one useful point. Send the message when they connect. Follow up twice, [48 hours] apart, then stop. Log every step in [CRM] and show me replies each day.
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
An AI dialer made 10,713 calls for a Sydney agency and live-moved 141 warm sellers to the owner.
Try it · based on the source's workflow
Call seller prospects for [agency] in [area]. Use local facts only from [sources]: [school data, transport, building history, recent sales]. Ask early: are you looking to buy or sell your property? If the person is a warm seller, move the call live to [agent name] at once. Tag anyone who asks for a check-in in [six months] for the next campaign. Log every call result in [system].
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
A four-hotel group moved guest answers from a 6% bot rate to 93% in one inbox.
Try it · based on the source's workflow
Answer guest messages for [hotel] on [channels]. Use only our own material: [FAQs, property guides, menus, local tips]. Answer in the guest's language. Cover [check-in, parking, restaurant hours, local tips]. If a question is outside this material or the guest is upset, pass the full thread to [staff contact] at once. Log every talk in [inbox].
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
A 14-room Tuscany villa's WhatsApp concierge took 70% of repeat questions, and experience sales rose 30%.
Try it · based on the source's workflow
Be the WhatsApp concierge for [property]. Answer from this material only: [rooms and rates, stay rules, restaurant hours, events, services]. Reply in the guest's language at once. Offer [added services] when they fit the stay. Any guest can ask for a person at any time. Pass any question outside this material to [team member] with the full thread. Show me all live talks in [dashboard].
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
A Guelph real estate agency filtered spam with a chatbot, and monthly bookings rose from one or two to five or seven.
Try it · based on the source's workflow
Be the website chat agent for [agency]. Filter out spam, link requests, and fake inquiries. Answer general questions from [services and area material]. For a real lead, collect name, phone, and what they need, and offer a booking call from [calendar]. Send the lead to [team inbox] with the full thread. Match our voice: [examples]. Show me each day's leads and filtered spam.
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
One operator runs five firms with no staff on a $48-a-month stack, and says it saves 30 to 40 hours a week.
Try it · based on the source's workflow
Run the daily system for [businesses]. At [time], pull new posts from [channels] and draft one [post] in my voice. Send it to me for review before it posts. Watch [forums] for posts about [topics] and draft replies for my review. Enrich each new lead from [source]: scrape their site, judge the fit, and write a one-page brief. Log every task and flag anything odd the same day. Send nothing without my review.
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
Three agents took a SaaS firm's first reply from hours to under 30 seconds, and meetings per rep rose over 200%.
Try it · based on the source's workflow
Work new leads for [business] in three lanes. Lane one: reply to each lead in [channels] within [30 seconds], in their language, at any hour. Lane two: ask [discovery questions], check [CRM] history, score fit and intent, and route hot leads to the right [rep] in under [two minutes]. Lane three: draft follow-ups from [industry, size, pain points] for rep approval, book in [calendar], and send a prep brief [30 minutes] before the call. Never discuss [pricing or terms]: a person owns those.
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
A two-person support team moved AI resolution from about 1% to 85% in under two months.
Try it · based on the source's workflow
Answer support chats for [business] in open chat, with no menu tree. Use only [help material]. Catch any talk that is angry or complex early and pass it to [staff contact] with the full thread. Never give [advice class your firm must not give]. Log each talk with its result. Send me a daily count of talks handled, resolved, and passed to a person.
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
An app team's agent resolves about 480 tickets a month and writes its own help articles from past tickets.
Try it · based on the source's workflow
Answer support tickets for [product] inside [help desk]. Use only [site material, files, past tickets]. Reply with a short delay, not at once. Each week, read the tickets staff solved, compare your draft to the real reply, and draft help articles for the gaps. Sort each ticket by class. Never reply to [sensitive classes]: send those to [staff contact] at once. Report tickets handled, resolved, and hours saved each week.
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
A helpdesk firm's AI answered 70 of 104 tickets in one week with one reply and no human touch.
Try it · based on the source's workflow
Answer new support tickets for [business]. For each ticket, search [knowledge base, docs, past tickets] first. Use [billing lookup] for order and account questions and [code or expert lookup] for hard technical questions. Send the first reply in our voice from these examples: [paste]. Never [refund, delete an account, or cancel a subscription]: those need a person. Log every reply for my review at [time].
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
A fashion brand's chat agent rose from 50% to 85% auto-resolved, with replies in three to six seconds.
Try it · based on the source's workflow
Answer shopper chats for [brand]. Train on [past chats and tickets]. For order, return, and refund questions, pull the real data from [order system] before you reply. Reply in [three to six seconds] in our voice: [examples]. If your confidence is low or the shopper is unhappy, pass the talk to [team] with a summary and the right department named. Log every talk.
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
A Texas surgery practice caught 41 new patients from missed calls in a 30-day pilot, worth $81,000 in production.
Try it · based on the source's workflow
When a call to [practice] is missed, text the caller at once from [agent name]. Hold a natural talk: gather [name, need, insurance, wanted times]. Send the full thread to [staff email] at once and add it to the [morning summary]. A staff member finishes each booking, so collect full facts and never confirm a time yourself. Flag any urgent case to [contact] at once.
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
A solo ghostwriter's morning prompt scans leads, drafts ten cold emails, and flags account problems for about $15 a month.
Try it · based on the source's workflow
Run my morning brief for [business]. Scan [email] for new leads, client news, and problems from the last [24 hours]. Check [CRM] for high-value contacts I have not reached. Search the web for one useful fact on my top [three] leads. Draft a cold email for each in my voice from these examples: [paste]. Update [dashboard] with the brief: leads, drafts, problems, and what needs me today. Send nothing.
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
A contactless hotel's agent resolves 90% of guest questions alone, and upsells added $20,000 to $30,000.
Try it · based on the source's workflow
Answer customer messages for [hotel]. Use only [policies, check-in steps, amenities, access rules]. For access problems, pull the customer's real booking code from [system] and send it. Offer [upsells] at the right stage of the stay. If a question is outside this material, pass the full thread to [team contact]. Log every talk in [inbox].
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
An agency owner built 13 single-job agents in one day, on the same $100-a-month plan, and revenue did not move yet.
Try it · based on the source's workflow
Build single-job agents for [business], one job each: [forum scan for buyer complaints, morning brief from all agent logs, lead finder with draft emails, content drafts from my notes]. Each agent runs on [schedule], logs every output to [board], and sends me a file at [time]. Nothing reaches a client without my review. Start with one lane, show me a test run, then add the next.
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
A freelance developer's nine small agents save about three hours a week and cost about zero on free tiers.
The prompt · from the source
Analyze this Reddit post. Is it a qualified lead for an AI automation service?
Title: {post['title']}
Text: {post.get('selftext', '')[:300]}
Return JSON: {"qualified": bool, "score": 0-10, "response_message": "2-3 line natural reply"}