June 12, 2026
It's Friday, and that means new updates from Invent. This week's batch is a big one: your team gets an AI co-writer in every reply, one-tap canned replies, follow-up rules in plain English, per-contact AI control, a real computer inside your personal chats, and a way to duplicate them when you need to branch.
We love hearing from business owners and agencies about how they keep scaling their businesses with Invent, and this week's updates buy your team time back: faster, on-brand replies on every customer message, and real finished work from every personal chat.
The Highlights
Assisted Human Replies: an AI co-writer in every text box
When you or your team takes over a conversation, the Assistant helps you write the reply instead of starting from scratch. Tap the magic wand and pick what you need: Polish to fix typos and tighten the message, More formal or Friendlier to shift the tone, Shorter to trim it down, or Draft a Reply to start from a fresh suggestion based on the full conversation context.
Tap the magic wand in the inbox to polish, reshape, or fully draft your reply from the conversation context.
Stop rewriting the same message ten times. Send it once, polished. Save time, reply better and faster.
Assisted Human Replies also lives in your personal chat, where the same magic wand polishes and expands your prompts before you send them. Type a quick idea like "market research for beauty salons in LA", tap Polish, and you get a richer prompt back with the audience, the goal, and a structured list of what you want included.
Same magic wand on the personal chat homepage: a quick prompt on the left, a polished, fully structured prompt on the right.
Like having a copywriter behind every reply and every prompt, without hiring one.
Canned Replies: save once, send in a tap
Save the replies you send all day and pull them up in one tap. Quick ones like "Yes, we deliver Friday" or "Our hours are 9 to 5." Or the longer ones: your full booking confirmation with policy, your pricing breakdown with add-ons, the refund process explainer, the onboarding welcome with next steps, the apology-plus-escalation message for tough conversations.
Canned Replies live inside the magic wand menu in the inbox, ready to send in one tap.
Every reply stays on-brand, your team answers faster, and new teammates ramp up without memorizing your playbook.
You will find Canned Replies under the magic wand menu in the Inbox. Save the messages you write fifty times a week, send them in one tap, and get back to running the business.
Follow-up Rules: tell your Assistant how to follow up
You can now tell your Assistant in plain English how to handle quiet conversations: which hours to follow up in, how many times to nudge, what keywords should stop or pause the follow-up, and what tone to use. Your playbook, your rules.
Write your follow-up rules in plain English, and your Assistant follows them on every quiet conversation.
A few example rules that work well:
- Hours: "Only follow up between 9:00 and 18:00, never on weekends."
- Frequency: "Follow up at most twice, then close."
- Stop keywords: "Stop after a clear no, or if the customer mentions a competitor."
- Tone: "Keep follow-ups short and warm, never pushy."
Find Follow-up Rules in your Assistant Settings, under Capabilities, Auto Follow-ups.
Turn off AI replies per contact
You can now switch off AI replies for a specific contact. Set it once on the contact, and your assistant stays out of every conversation with that person. You decide exactly who always gets a human.
One menu tap on the contact, and the Assistant stays out of every conversation with that person.
Perfect for:
- VIP and high-value clients you always handle personally.
- Human-only by default: one setting on the contact instead of toggling every new conversation.
- Personal contacts: friends, family, or your own number never get an automated reply.
It pairs with Human-only mode from #020: together they let you decide where AI replies and where it never does.
Your chat now has a real computer
Your chat now has a real computer behind it. Ask for something, and you get the finished file back: a chart as an image, a document ready to download, code that actually ran. In the demo below we asked for a chart of the top 5 energy drinks in the US for the past year. The chat searched the sources, wrote and ran the code, and handed back the PNG, ready to download.
Ask for a chart, get the file: the chat searches, writes and runs the code, and hands back the finished PNG ready to download.
What's under the hood:
- Runs real code: Python, TypeScript, and JavaScript in a secure sandbox.
- A workspace per chat: files live in an isolated filesystem, so every chat keeps its own work and projects never mix.
- You set the boundaries: network access is your call. Full access, package installs only, or completely off.
It's live today in your personal chats at useinvent.com. Open the composer menu and switch on Computer. And it scales from here: ad creation, document editing, website building, and end-to-end API workflows are all on the way.
Duplicate Chats: keep the original, explore the copy
Duplicate Chats starts a new chat with a copy of every message up to that point, linked files included. The original stays untouched.
Duplicate a chat to branch the conversation; the original thread stays exactly as it was.
Two ways you'll use it:
- Explore a different direction without losing the thread you've built.
- Reuse a great setup: a chat with the right context becomes a starting template you can duplicate per client, per project, per idea.
Where Duplicate Chats works:
- Personal chats (your chats at useinvent.com): yes.
- Assistant chats (Inbox and Playground): no.
- Org Chats: yes. These are new org-wide chats that help you manage assistants, conversations, and other org tasks, arriving in the coming weeks.
Duplicate Chats is handy in personal chats today, and it gets really powerful when Org Chats land.
Why it matters
Replies are where most of the day goes. Assisted Human Replies and Canned Replies cut the rewriting loop and lift the quality of every message your team sends. Follow-up Rules put the AI's behavior on rails you set yourself, and per-contact AI control keeps the relationships you choose fully human. On the personal chat side, your chat's new computer closes the gap between asking and getting finished work, and Duplicate Chats protects the context you've built so experimenting stops being risky. Less typing, more on-brand answers, more finished work, and full say over where AI shows up.
Weekend reading: real computing inside the chat is our take on AI that hands you the result. These three explain the shift:
- AI Agent vs Chatbot: What's the Difference for Your Business?
- What Is Agentic AI? A Business Owner's Guide
- The 4-Layer Anatomy of an AI Business Agent
New to Invent? Build your first AI Assistant in minutes with our docs and step-by-step guides.
With love,
Invent Team
Conversational AI for Business Growth
