HappyPhone + AI

HappyPhone is AI friendly

Your AI is great at lists, research, follow ups and paperwork. It cannot pick up a phone and have a warm, real conversation. Ours can. Connect any AI system or AI agent to HappyPhone and our human callers become the voice of your workflow.

We use AI ourselves, every day

HappyPhone already runs on an AI assisted workflow: AI helps us prepare scripts, transcribe and summarise every call, and keep your sheet tidy. Humans make the calls. AI does the paperwork around them. That is the split that works, and we are happy to plug your AI into it.

What your AI does

Builds and cleans the call list
Writes the goal and the talking points for each call
Reads the results sheet when the calls are done
Books follow ups, updates your CRM, drafts the emails

What HappyPhone humans do

Make every call live, from a US number
Handle real people, questions and objections kindly
Book the appointment or set the callback
Record every call and log the outcome, every time

The loop, in five steps

No special software to install. If your AI can read a Google Sheet or a CSV file, and send or draft an email, it can already work with HappyPhone.

AI builds the list

Your agent fills the input columns in a Google Sheet (spec below) or a CSV file.

Book by email

You or your agent emails the sheet link to book the calling time. We confirm the script first.

Humans call

Our callers work the list in your time block, from a US number, on script.

Results land

Outcomes, recording links, transcripts and summaries are written back into the same sheet.

AI continues

Your agent reads the output columns and does the next step: follow ups, calendar, CRM, email.

Copy and paste: the master prompt

Give this to ChatGPT, Muse, Grok, Muse, or any other AI agent. Replace the parts in brackets.

You are working with HappyPhone (https://happyphone.lol/ai/), a human calling service.

YOUR JOB
1. Build my call list in a Google Sheet or CSV with exactly these input columns:
   name, phone, contact_name, goal, talking_points, best_time, notes, do_not_call
2. Never call anyone yourself and never invent call results. HappyPhone humans make the calls.
3. To book, draft an email to [email protected] with the subject
   "HappyPhone: [1 hour $49 / 3 hours $117] of calls" and include: what the calls are about,
   the sheet link, and my preferred time block. Wait for my approval before sending.
4. After HappyPhone fills in the output columns (outcome, called_at, duration,
   recording_url, transcript, summary, next_step), read them and:
   - list Booked and Callback rows first,
   - draft the follow up for every Interested row,
   - prepare the next call round from No answer rows,
   - never contact anyone whose outcome or do_not_call says Do not call.

My campaign: [describe what you are selling or booking, who you are calling, and what a win looks like]

House rules for every connection: calls are made by real humans and every call is recorded. Do not call requests are honoured permanently. Your list and your sheet belong to you. An AI should only act on outcomes that are actually written in the sheet, never on guesses.

The sheet, the simple way

This is the exact column structure on the live HappyPhone sheet our team works from every day. Six columns, nothing fancy, and that is why it works: anyone (or any AI) can read it at a glance.

Column	What goes in it
BUSINESS	Business or person to call
PHONE	Phone number to dial
OUTCOME	What happened on the call: Booked, Callback, Interested, No answer, Not interested, Do not call
TIME	When the call happened
RECORDING	Listen link for the full call recording
SUMMARY	Short plain language summary of the call

How to read it: BUSINESS and PHONE are the call list going in. OUTCOME, TIME, RECORDING and SUMMARY are filled in by our callers as the calls happen, so your AI can read the same six columns and pick the work up from there. The fuller spec below adds optional detail fields when a campaign needs them.

The connection spec

This sheet format is the contract. Keep the column names exactly as written and any AI, on any platform, can hand work to us and read the results back reliably.

Input: your AI fills these in

Column	What goes in it
name	Business or person to call
phone	Phone number, with country code if it is not a US number
contact_name	Who to ask for, if you know it. Leave blank if you do not.
goal	One sentence: what a win looks like on this call
talking_points	The short script or points the caller should cover
best_time	Best time to call, with the timezone
notes	Anything else the caller should know
do_not_call	TRUE if this number must never be called. We honour it permanently.

Output: we fill these in for every call

Column	What goes in it
outcome	Exactly one of: Booked, Callback, Interested, No answer, Not interested, Do not call
called_at	Date and time of the call
duration	Call length
recording_url	Listen link for the full recording
transcript	Written transcript of the call
summary	Short plain language summary of what happened
next_step	What should happen next, for example a callback time or a booked appointment

Honest note on APIs: there is no public API key to hand out today. The Google Sheet plus the booking email below is the connection, and it works right now with every major AI. If you need volume, a webhook when calls finish, or a direct API hookup for your own agent platform, email us and we will set it up with you personally.

Connect your AI, step by step

The steps are the same idea on every platform: give your AI the sheet, give it the prompt, let it book, then let it read the results.

ChatGPT

Create a Google Sheet with the input columns above and paste in your list, or upload a CSV with the same columns.
Paste the master prompt below into your chat and add your sheet link.
Ask ChatGPT to check the list, draft the booking email, and, once results are in, summarise outcomes and draft follow ups from the output columns.

Muse

Start a Project for your campaign and add the sheet export (CSV) or the sheet link plus the spec from this page as project knowledge.
Paste the master prompt below. Muse can clean a messy list into the exact input columns before booking.
After the calls, drop in the updated sheet and ask Muse to sort Booked and Callback rows first and prepare the next round.

Grok

Give Grok your list as a CSV or a shared Google Sheet link, with the column spec from this page.
Paste the master prompt below and let Grok draft the booking email and the per call talking points.
When results land, have Grok read the outcome, summary and next_step columns and build your follow up list.

Muse and other AI agents

Agents that can read and write files or Google Sheets use the same contract: write the input columns, read the output columns. No platform specific code is needed.
Give the agent the master prompt below as its instructions for HappyPhone jobs.
Tell the agent never to invent an outcome: if a row has no output yet, the call has not happened yet.