What is Sidekick?

Sidekick is the conversation-intelligence layer that runs over every conversation in Atender. It reads what’s being said in real time, understands it, and routes that understanding to whoever — or whatever — is on the other end. For human agents, Sidekick shows up as a slide-out panel with summaries, suggestions, and matching knowledge. For the system, it works in the background — tagging conversations, detecting structured data, populating custom fields, spotting incident matches.

Sidekick assists. Agent Stacks act. Agent Stacks handle conversations autonomously; Sidekick understands every conversation and surfaces that understanding wherever it’s useful — to the agent in the panel, to the system in the background, and increasingly to Agent Stacks themselves.

Three kinds of understanding

Every time a customer message arrives, Sidekick reads the conversation and extracts:

• Topic — What the customer is asking about, what’s already been tried — Conversation summary in the panel
• Entities — Structured data inside messages — order IDs, serial numbers, emails, phones, custom patterns — Inline highlights + auto-saved to custom fields
• Patterns — Cross-conversation relationships — incident matches, multiple open conversations from the same contact — Hint cards above the panel

That understanding is the same regardless of who’s handling the conversation. What changes is how it’s used.

For agents — the panel

When an agent opens a conversation, Sidekick appears as a slide-out panel containing a stack of modules. Each module shows a different kind of help: a quick summary, a suggested response, matching Knowledge Base articles, internal Handbook procedures, relevant links.

Modules are configurable per tenant — enable, disable, reorder via drag-and-drop in Settings → Sidekick. See the panel modules reference for what each module does.

For agents — the hint cards

Two contextual cards can pop above the modules when something material is happening:

• Possible incident match — when a customer describes a problem matching an active incident, Sidekick offers to link the conversation and subscribe the customer to updates. See Link a conversation to an incident from the panel.
• Multiple active conversations — when the same contact has other open conversations, Sidekick flags them with a count, links, and a quick-merge action.

For the system — background workers

Most of what Sidekick does is invisible. Three background workers run on every conversation regardless of whether a human is reading the panel:

1. Detection rules — pattern-matching that extracts structured data (order IDs, serial numbers, anything you teach it) and auto-saves to custom fields. See Detection rule types and Create a detection rule.
2. Auto-tagging — applies tags whose descriptions semantically match the conversation. Off by default — opt tags in via Tag Management and enable the global toggle in Settings → Sidekick.
3. Incident discovery — runs the matching that powers the incident-match hint card. Toggleable in settings.

What you tune

Sidekick’s behaviour is controlled from Settings → Sidekick, organized into four tabs:

• General — the master switch, incident discovery toggle, inline-highlight toggle, auto-save toggle, confidence threshold.
• Panel layout — module ordering and visibility.
• Detection rules — create, edit, reorder, test rules. Map to custom fields.
• Auto-tagging — enable the worker, set confidence threshold, set max-candidates cap.

Where to start

• Tour the panel: Panel modules reference
• Set it up: Configure the Sidekick panel
• Extract data: Create a detection rule
• Auto-tag conversations: Enable AI auto-tagging on a tag