Question 1

What triggers can start an automation rule?

Accepted Answer

Twenty-plus trigger types including new conversation, message received, AI handoff, tag added, SLA breach risk, priority change, customer tier match, time since last message, and webhook events. Triggers combine with if/else conditions to express almost any workflow.

Question 2

How do we escalate a conversation that is running out of time?

Accepted Answer

With a rule. As a conversation crosses At risk, Urgent and then Breach, each step fires the sla.warning or sla.breached event, and a rule that listens for it does the escalating — reassign to a team lead, raise the priority, post to a channel. You decide what escalation means. The policies themselves are set up on the CSAT & SLA page.

Question 3

What are Cases and how do they differ from conversations?

Accepted Answer

Cases are for work that spans days — disputes, returns, onboarding — with custom stages, Kanban tracking, and due-date aging. Each case supports forms, file uploads, and e-signatures, and links bidirectionally to the conversations that spawned it.

Question 4

Do I need developers to set up automation?

Accepted Answer

No. The visual multi-step editor builds every rule without code. If you do need to reach external systems, webhook actions and API calls are first-class. For AI agents the same job is done by Capabilities, where you point Atender at one of your API endpoints and publish it as an action an agent may run.

Question 5

What happens when multiple rules match the same conversation?

Accepted Answer

Rules execute in a defined order you control, with explicit stop conditions to prevent runaway chains. Each rule can flag whether subsequent rules should continue evaluating, so conflicts are resolved by your configured priority — not by chance.

Question 6

Can we test a rule before turning it on in production?

Accepted Answer

Yes. Every rule has a dry-run mode that evaluates against live conversations and logs what would have happened without taking action. Sandbox mode lets you run the full rule chain against a copy of production data before publishing.

Question 7

Are rule changes versioned, and can we roll back?

Accepted Answer

Yes. Every edit is versioned with author, timestamp, and diff. Roll back to any prior version in one click, and compare versions side by side to see exactly what changed when a workflow stops behaving as expected.

Question 8

How reliable are webhook actions — what about retries?

Accepted Answer

Webhook actions retry with exponential backoff on transient failures (network errors, 5xx responses), with configurable retry counts and dead-letter logging. Failed webhooks surface in the rule's run history so you can investigate without digging through logs.

Question 9

Can automation rules use AI mid-workflow?

Accepted Answer

Yes. LLM nodes can classify, summarize, extract entities, or generate responses as steps inside a rule — so an automation can read an inbound email, decide what it's about, and route or reply accordingly. AI agents go further: a capability lets an agent call one of your own API endpoints, with an action tier deciding how much proof the customer needs first.

Question 10

Are there limits on rule complexity?

Accepted Answer

Practical limits, not hard caps. Rules can chain dozens of conditions and actions across branches. The visual editor stays readable as workflows grow, and shared logic is best kept in your own system and reached from a webhook action or an API call.

Set the rules. Let the platform run.

What's in Automation

If this, then that — and more

Rules that fire before a deadline does

Track work that takes time

Time-aware automation

Conversations

AI Agents

Analytics

Set the rules, let the platform run.

See Atender in action