Auto Follow-ups

Make your agent reach back out — automatically

Most leads don't close on the first message. Auto Follow-ups let your Kronos agent re-engage a customer after silence — to nudge a decision, recover a stalled chat, or just say "hey, still interested?" — without you lifting a finger.

TL;DR

Open your agent → Engagement tab → Auto Follow-ups.
Pick a mode: Off, Agentic (recommended), or Guided.
Set Maximum attempts (1–10) and Inactivity timeout (1–30 days).
Save. Your agent will follow up on its own from now on.

Where to find it

Open the agent you want to configure (AI Agents → your agent), click the Engagement tab on the left sidebar, and you'll land on the Auto Follow-ups panel:

Engagement tab inside an AI agent showing the three follow-up modes (Off, Agentic, Guided) and the Maximum attempts / Inactivity timeout fields — The Engagement tab. Mode picker on top, retry knobs underneath.

The three modes

Each mode answers a single question: who decides when (and what) to send next?

Off

No automatic follow-ups

Who decides: Nobody.
What it does: The agent only replies when the customer messages first. A stalled chat stays stalled forever.
Best for: Channels where unsolicited messages would feel pushy, or while you're still tuning your agent and don't want it sending anything you haven't reviewed.

Recommended

Agentic

The AI decides when and what to say

Who decides: Your agent.
What it does: After each conversation turn, your agent reads the chat and decides for itself whether — and when — to follow up. It picks both the timing (anywhere from seconds to days) and the message content based on context.
Best for: Sales and engagement. The agent can be aggressive when a customer says "let me think about it for a few hours" and patient when they say "next week".

Guided

You give the playbook, the AI delivers it

Who decides: You + your agent.
What it does: You provide a follow-up plan in your agent's instructions (e.g. "If they don't reply in 2h, ask if they need help with pricing"). The agent decides timing within those rails and adapts the tone to the conversation.
Best for: Teams that already have a proven outreach script and want consistency, but still want the agent to phrase things naturally.

Maximum attempts & Inactivity timeout

These two numbers exist to keep your agent polite and stop a stale chat from becoming a robocall.

Maximum attempts

1 – 10

How many follow-ups the agent will send before giving up. Each customer reply resets the counter — so a back-and-forth that goes silent again is a fresh runway.

e.g. Set to 3: after 3 consecutive un-replied follow-ups, the agent stops trying.

Inactivity timeout

1 – 30 days

Days of silence after which the agent stops following up entirely, even if it hasn't hit the maximum-attempts limit. Use it as a hard floor on how long a conversation can stay 'open'.

e.g. Set to 7: if a week passes with no customer reply, the agent moves on.

Whichever limit hits first wins. Three attempts in two days will stop, and so will one attempt that's been waiting 30 days. The customer can always restart the conversation by replying.

What actually happens

1Customer messages your agent on WhatsApp / Instagram / Messenger.
2Agent replies. If Off, nothing else happens until the customer messages again.
3On Agentic or Guided, the agent privately decides whether a follow-up makes sense and, if so, schedules it for some time in the future (could be 30 seconds, could be 3 days).
4When that time arrives, Kronos wakes the agent up, it reads the conversation again, and sends a fresh, context-aware message — not a canned template.
5If the customer replies, the cycle restarts and the attempt counter resets. If not, the agent may schedule another follow-up (up to Maximum attempts).

Which mode should I pick?

If you...	Use
...are still testing the agent and don't want it sending anything autonomously	Off
...sell a service or product where re-engagement matters and you trust the AI to choose timing	Agentic
...have a proven outreach playbook (e.g. "day 1 nudge, day 3 offer, day 7 last call") and want the AI to follow it	Guided
...handle support only — customers always start the chat and don't need a nudge	Off

Practical tips

Start conservative. 3 attempts & a 7-day window is a safe default. Tighten only if data shows customers want more nudges.
Mention follow-up boundaries in your agent's instructions. e.g. "Never follow up at night — wait until 9am the customer's local time." Agentic mode will honor that.
Combine with email notifications. A "customer went cold" objective + Email = you get notified when the agent gives up, so you can step in personally.
Test in the Test Chat first. The Test Chat does NOT trigger real follow-ups (it's an ephemeral playground), so use it to validate your agent's tone, then watch real conversations to tune the knobs.
WhatsApp 24-hour window. WhatsApp only allows free-form messages within 24h of the customer's last reply. Outside that window, follow-ups must use a pre-approved template.

Open your AI agents

Edit any agent and pick the Engagement tab.

Email notifications

Get notified when the agent closes (or gives up on) a follow-up loop.

WhatsApp templates

Required for follow-ups outside the 24h customer reply window.

FAQ

Will the agent spam my customers?

No. Both Maximum attempts and Inactivity timeout exist as hard ceilings. On top of that, in Agentic mode the agent itself is biased toward fewer, higher-quality nudges — it won't schedule a follow-up just because it can.

What does "the agent decides when" actually look like?

The agent picks a delay anywhere from a few seconds (rare, used for "hey are you still there?" mid-conversation) to several days (waiting for a quote to mature). It also writes a short note to itself about why it's following up, which it uses when the moment arrives so the message stays on-topic.

Does the customer's reply reset the counter?

Yes. Any new customer message ends the current follow-up loop and starts fresh. If the conversation stalls again later, the agent gets a brand-new budget of attempts.

Can I disable follow-ups for a single conversation?

Pick Off mode globally, or simply take over the chat manually from the inbox — the agent will not auto-follow-up on conversations a human has actively replied to recently.

Do follow-ups work in the Test Chat?

No — the Test Chat is intentionally ephemeral, so no real timer-based follow-ups fire. Connect a real channel and message the agent from a personal account to see follow-ups in action.

What if my agent crashes or runs out of credits?

If a scheduled follow-up can't be delivered (provider error, credit issue, etc.) Kronos puts the conversation back in line and notifies you in the bell, so the loop isn't silently lost. Re-runs respect the same attempt count.

Can I see the agent's reasoning for a follow-up?

Yes. Open the conversation in the Inbox — the activity panel next to the chat shows the agent's notes, including why it scheduled (or canceled) a follow-up.

Does Guided mode require a specific format in my instructions?

No, plain English (or Spanish) works. Just describe the playbook you want — "wait 4h, then ask if they want pricing; if no reply in a day, send the case study" — and the agent will execute it while adapting the tone.