AI Agent: reply in the thread or send a new email?

You have configured your AI agent to follow up with a client who has not responded to your quote. The email is sent. But where does it land? In the discussion thread where you sent the initial quote, or in a separate new email? This technical question has very concrete consequences on your recipient's reaction. Let's see what happens in each case.

IWhy does this question come up?

A person replying to an email clicks "Reply" without thinking. The email stays in the same thread. But an AI agent can be configured differently. It can create a new message, with a new subject, that arrives separately in the recipient's inbox. Both approaches are possible, and each has its logic. The choice depends on what you want to accomplish.

IIReplying in the thread: when it's the right choice

The reply maintains continuity. Your recipient immediately finds the context: the quote sent, their initial question, your first response. It's natural, fluid, and resembles what a human would do. For a quote follow-up, for example, the client immediately sees what it's about. They don't need to search their inbox. The reply works when the relationship is already established and the subject is the same.

IIISending a new email: when it's preferable

A new email creates an intentional break. It grabs attention because it appears as a distinct message, with its own subject line. It's useful when the topic changes, when you contact someone for the first time, or when the previous thread has become too long and confusing. In prospecting, for example, each outreach is a new email: there's no existing conversation.

The new email is also relevant when you want to mark a milestone. An accounting firm sending a reminder of tax documents can choose a new email so the client doesn't confuse it with regular exchanges. The email subject then becomes a clear signal: "Missing documents for your tax return."

IVHow to choose the structure of the agent email?

VConcrete examples by situation

VIHow to configure your AI agent?

When you create an agent on ABYS, you describe its mission in your own words. That's where you specify the expected behavior. You can write: "For quote follow-ups, reply in the existing conversation thread. For prospecting, send a new email with a personalized subject." The agent follows these instructions. If you don't specify anything, it will choose based on context, but it's better to be explicit.

VIIWhat your recipient sees

A reply arrives in the existing thread. On Gmail, Outlook, or most email clients, it appears below the previous messages. The subject remains the same, preceded by "Re:". A new email appears separately, with its own subject, like a distinct conversation. For your recipient, the difference is immediate: either they recognize a continuation, or they discover a new subject.

VIIIA common pitfall: mixing the two

Some configure their agent without thinking about it, and the result is incoherent. One follow-up goes out as a new email, then the next one as a reply, then another new email. The recipient gets confused. They have three conversations about the same quote. To avoid this, choose a clear rule and stick to it. If your agent handles multiple types of missions, specify the behavior for each one.