July 24, 2026
TL;DR
- AI Routing: assistants transfer conversations to specific Roles or Members via natural-language rules you set.
- Template Replies: reply to WhatsApp conversations outside the 24-hour window with approved templates.
- Delivery Pace for Broadcasts: spread a send over 1 hour to 7 days at the rate you pick.
- 4 new Gemini models: Google's new Flash 3.6 and Flash Lite 3.5 families (plus Reasoning variants) live in the picker.
It's Friday, and that means fresh updates from Invent. This week is all about control: who handles what, when your broadcasts go out, and reopening the WhatsApp window when you need to.
Introducing AI Routing 🎉
Your Assistant can now transfer any conversation to a specific Role or a specific Agent, and it decides based on natural-language rules you set per member.
Write what each member handles ("billing disputes and refunds, English and Spanish"), and the Assistant routes billing questions to that person the moment they come in. If everyone on your team handles the same things, skip the instructions and it routes by role.
- Transfer to Roles or Members, not just to a generic "any human".
- Natural-language routing instructions per member, so the Assistant knows what "Jorge handles" without you writing rules.
- Every conversation lands with the person who owns it, no more "who wants this?" pings.
Set it up under Settings → AI Routing or under Members.
Every teammate can have AI Routing turned on, with their own routing instructions attached.
Write what each person handles in plain English. The Assistant reads it and routes accordingly.
Template Replies for WhatsApp, even after 24 hours
WhatsApp closes conversations after 24 hours of silence. You used to lose that thread unless a customer messaged back first, or you paid for a template send from Meta Manager.
Not anymore. Template Replies let you re-engage any out-of-window conversation with an approved WhatsApp template, right from the chat. Pick a template, drop in variables and a header image, and send. The conversation reopens.
Great for follow-ups after a demo, campaign nudges, abandoned bookings, and any thread that went quiet.
When a conversation is outside the 24-hour window, Invent surfaces a Reply with template button right in the chat.
Pick an approved template, drop in variables and a header image, preview it live, and send.
Delivery Pace for Broadcasts
Sending 500 WhatsApp messages in a single burst can trip provider limits and pile every reply on your team at once.
Now you control the pace. Spread a broadcast over 1 hour, 3 hours, 12 hours, a day, or up to a week. Invent drips them at the rate you pick, and your team handles the replies at a human pace.
- Preset options: 1h / 3h / 12h / 1d / 3d / 7d.
- Recommended pace highlighted based on your list size.
- Fewer angry customers waiting, and a happier team.
Set the pace from the Send Broadcast panel when you send.
Delivery Pace: pick how long the send takes to complete, from a minute to a week. Invent recommends a pace based on your list size.
4 new Gemini models in your picker
Google shipped its new Gemini 3.5 and 3.6 families this week, and all 4 variants are already in Invent:
- Gemini Flash 3.6: higher speed, strong at code generation and execution.
- Gemini Flash 3.6 Reasoning: extended thinking for the tough calls.
- Gemini Flash Lite 3.5: Google's cost-efficient GA model, high-volume agentic tasks and simple data processing.
- Gemini Flash Lite 3.5 Reasoning: extended thinking, cost-efficient.
Pick any from the model selector on your Assistant, or leave it on Auto and let the picker route to the right model per task.
Gemini Flash Lite 3.5: Google's cost-efficient model, tuned for high-volume agentic tasks.
Gemini Flash 3.6: Google's faster model, tuned for code generation and agentic execution.
Why it matters
Every update in this batch reinforces the same theme: more control. Control over who on your team handles each conversation. Control over reopening WhatsApp windows past 24 hours. Control over the pace your broadcasts hit customers. And more choice over which model runs your Assistant. The AI does the work, you set the rules.
With love,
The Invent Team
Conversational AI for Business Growth
