September 25, 2026
TL;DR:
- Invent Composer starts WhatsApp conversations
Type any phone number in the Inbox search bar, even one not in your contacts. Invent creates the contact, and the Composer opens the WhatsApp conversation with an approved template, the way Meta requires. - GPT-6 Luna in the model picker
OpenAI's new model for high-volume work, with a 1M-token context window and reasoning. Pick it per assistant, switch anytime. - Five days to Meta's WhatsApp deadline
Payment method on your WhatsApp Business Account by September 30, or replies stop on October 1.
Friday again. Last week we rebuilt the Composer. This week it stopped waiting for customers to write first.
Invent Composer starts WhatsApp conversations
Until now, a WhatsApp conversation in Invent started when the customer wrote to you. Now you can start it. Open your Inbox, type any phone number in the search bar, even one not in your contacts, and Invent creates the contact on the spot, and the Composer we rebuilt in #037 takes it from there.
Type the number in the Inbox search bar and Invent creates the contact on the spot.
The number you got from a lead form, a business card, or a missed call now has a WhatsApp conversation ten seconds later, from the same inbox your team already works in.
The first message is a template, as Meta requires
WhatsApp has a rule for businesses that write first: the opening message has to be a template Meta approved. Invent handles that for you. When the new conversation is on WhatsApp, the reply box switches from Reply to Template, you pick one from your approved list, and it goes out. Once the customer answers, the conversation continues like any other, with the AI assistant on if you want it.
On a new WhatsApp conversation the Composer asks for a template before anything else.
The Template tab lists every template on your account with its language and approval status, and a search box for when the list gets long.
Your approved templates, one tap away, so a new WhatsApp conversation starts inside Meta's rules.
New to templates, or not sure which category to use? Our WhatsApp templates guide covers approval and categories.
GPT-6 Luna joins the model picker
GPT-6 Luna is OpenAI's model for focused, high-volume work, released on September 22. It brings a 1M-token context window, reasoning support, and takes text, images, and documents. In the picker it scores top marks on speed, which makes it a strong fit for assistants that answer a lot of routine questions fast.
GPT-6 Luna: fast, 1M context, text, images, and documents in, at the low end of the cost scale.
As always, you control the model. Pick it for one assistant, keep another on Gemini or Claude, and switch anytime without touching your instructions.
A friendly WhatsApp reminder: 5 days left
If your WhatsApp is connected to Invent, you are on the API and this applies to you. Add a payment method to your WhatsApp Business Account by Wednesday, September 30 and your number keeps answering on October 1. Two minutes in Meta Business Suite, and our step-by-step guide shows every click.
The first 1,000 replies each month, per phone number, stay free. Every country's rate is in our WhatsApp API pricing guide. This is Meta's fee, billed by Meta, and it has nothing to do with your Invent plan.
Why it matters
Invent used to answer the conversations customers started. Now the Composer starts them too, with the WhatsApp rules built into the flow instead of left for you to remember. And GPT-6 Luna gives the busiest assistants a fast model with room for the whole conversation.
With love,
Invent Team
Conversational AI for Business Growth
