the AI suggests, your team decides.
heysteff is an AI platform for customer support and sales across WhatsApp, Instagram, Messenger, Gmail and Shopify; Steff is the AI agent that runs it. Not every team wants the AI replying on its own from day one, and with heysteff they don't have to: copilot mode lets Steff propose the reply and lets a person decide whether it goes out as-is, gets edited, or gets discarded.
How it works
In copilot mode, every time a message comes in, Steff drafts a suggested reply using the same context it would use to answer on its own: the conversation history, your knowledge base, and, where relevant, customer or order data. That suggestion shows up in the human agent's inbox, who can send it as-is with one click, edit it before sending, or discard it and write their own reply from scratch.
The result is that your team still replies faster — nobody starts from a blank page on every message — but nothing goes out without a person reviewing it, if that's how you want to run things.
1-click bot pause
You can pause Steff with one click at different levels: a single conversation, an entire channel (WhatsApp only, for example), a whole workspace, or globally across the whole account. That's useful both for sensitive cases where you'd rather only a human replies, and for moments where your team simply wants to take over temporarily.
"Echo" detection: Steff pauses itself if your team already replied
If someone on your team replies to a customer directly from the native Instagram or Messenger app — bypassing heysteff — the system detects that reply as an "echo" and pauses the bot on that conversation automatically. That avoids the worst-case scenario of a badly implemented copilot: the AI replying on top of what a human already answered, producing contradictory or duplicate messages in front of the customer.
Escalation with context
When a conversation escalates to a human — whether the agent decides to based on your rules and keywords, or someone on the team takes it over manually — heysteff builds a summary of what happened up to that point. Whoever picks up the conversation doesn't have to read the entire history from scratch to understand what the customer needs and what's already been said.
From copilot to autopilot, at your own pace
Copilot mode isn't a mandatory starting point or a ceiling — it's a level of trust you control. Many teams start by reviewing every suggestion, and as they see Steff answering well and consistently, they hand off entire channels or conversation types to autonomous mode, while keeping human oversight where it actually matters — complaints, big sales, VIP customers — through rule-based escalation.
FAQ
Can I start in 100% copilot mode, without the bot replying on its own?
Yes. Many teams start there: Steff suggests every reply and a person approves, edits or discards it before it sends. It's a natural way to build trust in the agent before letting it respond autonomously, and you can move to autopilot whenever you decide, gradually or all at once.
How do I stop the bot from replying to a customer my team is already handling?
You can pause the bot in one click per conversation, per channel, for the whole workspace, or globally. On top of that, if someone on your team replies directly from the native Instagram or Messenger app (bypassing heysteff), the system detects that "echo" and pauses the bot on that conversation automatically, so replies never duplicate or contradict each other.
Who on my team can see what?
heysteff has roles and permissions: you decide who can pause the bot, who sees which conversations, and who can escalate. Every relevant action is recorded in an audit log, so there's always a trail of who did what and when.
◆ next step
Start in copilot mode and move to autopilot once your team is ready.
Also relevant: MCP server, appointment scheduling, AI lead qualification, AI agent for ecommerce and modo copiloto (Spanish).