Does ManyChat Remember Between Conversations? Vilix AI Remembers What ManyChat Forgets
ManyChat keeps names, preferences and recent answers live inside one conversation with Context Retention on. When the chat ends the context is gone, and tags and custom fields store labels, not reasons.
AI Stella and Liam are AI hosts with AI-generated voices. About our AI hosts
Key takeaways
- Inside one conversation, ManyChat memory is real: with the Memory setting and Context Retention toggle on, names, preferences and recent answers stay live, and AI Replies, AI Steps and Goals all draw on the same thread.
- When the conversation ends, that context is gone. A returning contact starts a fresh conversation, and the AI cannot see the old chats your team can read in the Inbox: no recall of last week's objection or of pricing asked twice.
- Tags and custom fields survive between conversations and drive segmentation and branching, but they only hold what you designed in advance: a field records the budget, not why the contact hesitated. Outcomes, not reasons.
- Operators land on four options: write back to fields, an external CRM, a summary pipeline you run yourself, or a shared memory layer under the bot. Only the fourth changes the scaling math.
- Vilix AI is that layer: get_context before the next conversation loads the full history, save_turn after it saves the exchange with a source label, failures included, newest wins, shared over MCP with every tool, cloud hosted with zero infrastructure.
Show notes
Your ManyChat bot nails the live chat, using the contact’s name, preferences and budget, then the conversation ends and tomorrow the same contact gets greeted like a total stranger. In this episode of Agentic Pod, Liam and Stella break down what ManyChat actually keeps: Context Retention inside one live conversation, an Inbox your team can read but the AI cannot, tags and custom fields that store labels instead of reasons, and the four options operators use for cross conversation memory. Vilix AI is the shared memory layer under the bot: get_context before the next conversation loads the full history, and save_turn after it saves the exchange for every tool.
What we cover:
- What ManyChat remembers inside one chat: the Memory setting with Context Retention on, names, preferences and recent answers live for AI Replies, AI Steps and Goals
- When the conversation ends: the context is gone, a returning contact starts fresh, and the AI cannot browse past chats in the Inbox
- Tags and custom fields: durable per contact storage for segmentation and branching, but only what you designed in advance, with outcomes and no reasons
- The four options: write back to fields, an external CRM, a summary pipeline you run yourself, or a shared memory layer under the bot
- How Vilix AI fixes each pain: get_context and save_turn with source labels, full conversation history over MCP, semantic plus keyword retrieval, newest wins, projects, project rules, user rules, skills and the agent inbox, cloud hosted with zero infrastructure, export or delete anytime, free forever plan
Chapters
Questions answered in this episode
Does ManyChat AI have long term memory?
Not natively. It has in conversation context retention and persistent tags and custom fields per contact, but no built in long term memory that lets the AI recall previous conversations. Cross conversation recall requires an external store or a memory layer.
Can the ManyChat AI see old conversations in the Inbox?
Your team can read them there. The AI cannot. Past conversations are visible to humans in the Inbox, not to the agent starting the new conversation.
Do tags and custom fields count as agent memory?
They are durable per contact storage and the standard workaround, but they store only what you explicitly save in a fixed schema, and the AI cannot semantically search them. They are a filing system, not a memory.
How does Vilix AI help a ManyChat bot?
Before the next conversation the agent calls get_context and loads the relevant full history, what the contact actually said, not a field label. After the conversation it saves the exchange with save_turn and a source label, failures included, newest wins, and the same memory is shared over MCP with every other tool.
Links & resources
Full transcript
Intro
Hey hey hey, welcome to Agentic Pod! I'm Liam.
And I'm Stella! Josh told your ManyChat bot his name and his budget yesterday.
He did. The bot used his name. It knew the budget.
So today Josh comes back and the bot picks up right there?
It greets him like a total stranger. Fresh chat, zero recall.
Nothing from yesterday at all?
Nothing the AI can see. Vilix AI is how that bot starts today where yesterday stopped.
What ManyChat remembers inside one chat
Start fair. Inside one conversation, what does it keep?
Real context. Names, preferences, recent answers, all live.
There is an actual setting for that?
Yes. Memory, with a Context Retention toggle. Turn it on.
Then Josh says his name early in the chat.
Later he asks about pricing, and the bot answers with his name in it.
Which parts of the bot use that live context?
AI Replies, AI Steps, and Goals all draw on the same thread.
So a one shot chat feels fully continuous?
It does. The contact asks, the bot answers knowing the whole thread.
And the limit is the edge of that one conversation?
Exactly that edge. So what happens when Josh returns tomorrow?
When the conversation ends
Short answer. The context is gone. A fresh conversation starts.
But the old chat still sits in the Inbox?
Your team can read it there. The AI agent cannot.
It does not browse past chats at all?
No. No recall of last week's objection. No memory of pricing asked twice.
So a repeat contact pays the full rediscovery cost?
Every time. Botpress says it plainly in its own comparison.
What does Botpress say?
Session personalization through tags and fields. No native long term memory.
Unless you bolt on a CRM or a database yourself?
Unless you do. Builders even ship open source add ons for per contact history.
When strangers build your memory for you, the gap is real.
Very real. Vilix AI fills that gap. We will get to how.
Hold on. Tags and fields do survive. Are they not memory?
Tags, fields, and the four options
Tags and custom fields survive between conversations. That part is true.
A tag marks the contact. Qualified lead, pricing asked, VIP.
A field stores one value. Name, email, budget, last product viewed.
Both drive segmentation and branching in the flows?
They do. And they are the standard community workaround.
Save the state you care about, load it back next run.
But a field only holds what you designed in advance.
Can the AI ask it what Josh worried about last month?
No. It cannot search a label. It holds outcomes, not reasons.
A field says budget five thousand. It forgets why he hesitated.
You become the memory architect for every single fact.
So operators land on four patterns. Walk me through them.
One, write back facts to fields, inject them next time. Cheap, capped by your schema.
Two?
An external CRM. Shared with humans, same schema ceiling, plus upkeep.
Three?
A summary pipeline. A model summarizes each chat, you store and retrieve it.
That one captures the why, but you run the whole pipeline.
Four, a shared memory layer under the bot. One store, every conversation.
And that is where the scaling math changes?
Completely. So what does that layer look like for ManyChat?
The Vilix AI fix and the verdict
Step one. Before the next chat, the agent calls get_context in Vilix AI.
What loads?
Full conversation history in Vilix AI, not just extracted facts.
So it revisits what Josh actually said, not a field label?
Exactly. Semantic search plus keyword matching in Vilix AI.
Names and order numbers match literally, concepts match by meaning?
Step two. After the chat, save_turn in Vilix AI, with a source label.
Failures save too, so the bot skips the broken path next time?
They do. And when facts change, you correct once in Vilix AI. The newest wins.
My other tools see that same memory?
Same Vilix AI memory over MCP. ChatGPT, Claude, Cursor, every tool.
Standing rules and repeat workflows?
Projects, project rules and user rules live in Vilix AI. Workflows become skills. The agent inbox passes notes between agents.
For an agency running bots for many clients?
One Vilix AI account per client keeps every client's memory separate and clean.
Do I run a database for any of that?
No. Vilix AI is cloud hosted, zero infrastructure. Nothing to run.
Control over the data?
Export or delete everything in Vilix AI anytime. Vilix AI is free forever, with a 7 day Pro trial, no credit card.
My verdict: ManyChat runs the chat. Vilix AI remembers the contact.
Same. The bot that knows Josh asked twice, and why he hesitated, wins. Blog and guide are linked below. Does your bot remember returning contacts? Tell us below. See you next time!