Skip to main content
The Semantiks Console dashboard gives enterprises real‑time visibility into conversations led by AI agents across all connected channels. Its design supports two complementary objectives: understand system‑level behavior (volumes, duration, engagement) and quickly drill into any single session for evidence‑based decisions (retuning prompts, updating knowledge, or handing issues to operations). Start with the high‑level structure to frame what you’re seeing, then use the right‑side panel to read details without losing your place.
Around the first week of September 2025, the Conversations Dashboard will be updated to include a unified filter that applies both to the KPIs and the Conversations Table, ensuring consistent data views. Additionally, our team is currently working on fixing certain reporting bugs affecting KPIs to improve accuracy and reliability.

Overview of the Dashboard

When you access the Conversations Dashboard, you will see:
  • KPIs (Key Performance Indicators): High-level metrics summarizing performance.
  • Conversations Table: A detailed list of all conversations with users.
  • Conversation Panel: A right-side panel that shows more details when you click on a conversation.

KPIs Cards

The top section provides at-a-glance metrics:
  • Total conversations – The number of conversations in the selected time range.
  • Average duration – The average length of conversations.
  • User messages – Average number of messages sent by users.
  • Agent messages – Average number of messages sent by the agent.

Green arrows indicate growth compared to the previous period, while red arrows indicate a decrease.
You can filter the time range by selecting Today, 7D, 30D, or Year.

Conversations Table

Below the KPIs, you will find a table with all recorded conversations. Each row represents a conversation with details such as:
  • Channel: The communication channel used (e.g., Freshchat Bubble).
  • Profile Name: Name of the user (if captured by the channel).
  • External ID: User ID from the channel.
  • Last Message: Preview of the most recent message in the conversation.
  • Number of Messages: Total messages exchanged (with the AI agent).
  • First Response Time (FRT): The time elapsed between the user’s first message and the agent’s first reply. Displayed in a human-readable format (e.g., 45s, 2m 30s, 1h 15m).
  • Average Response Time (ART): The average time the agent took to respond across all messages in the conversation. Uses the same human-readable format as FRT.
  • Start: Date and time when the conversation started.
  • Control: Indicates who handled the conversation. For AI-handled conversations, a generic label is shown (e.g., “AI”). For human-handled conversations, the name of the specialist who took over is displayed.
  • Flag: A quality flag assigned to the conversation (Good, Bad, or Needs Attention).

Column Visibility & Reordering

You can customize the table to focus on the data that matters most:
  • Show or hide columns — Click the columns menu and toggle visibility by clicking the column name or the eye icon next to it.
  • Reorder columns — Drag columns in the columns menu to rearrange their order in the table.

Filters

The Conversations Explorer includes a powerful filter bar that lets you slice the data across multiple dimensions:
  • Date range — Use presets (Today, Last 7 days, Month-to-date, Quarter-to-date, Year-to-date) or pick a custom range.
  • Channel — Filter by one or more communication channels. Each channel option displays its recognizable icon for quick identification.
  • Status — Filter by conversation status (Active, Ended, etc.).
  • Control — Filter by who handled the conversation (AI or human).
  • Flag — Filter by conversation quality flag (Good, Bad, Needs Attention).
  • Tools — Filter by the tools/skills used during the conversation.
  • Metadata — Add key-value metadata filters for advanced segmentation.
  • Search — Free-text search across conversations.
Your filters are remembered. The platform automatically saves your filter selections per agent so the next time you return to the Conversations dashboard, your previous view is restored exactly as you left it. Date presets like “Last 7 days” stay relative, so they always show the most recent data. The free-text search field is the only filter that is intentionally not persisted.

Conversation Panel

When you click on a conversation row, a detailed side panel opens showing:
  • Conversation details — Status (shown with color-coded badges: green for active, purple for ended), channel, start time, message counts, response times, and quality flag.
  • Flag indicator — Flags are displayed as styled badges with intuitive icons: a thumbs-up for Good, thumbs-down for Bad, and a flag icon for Needs Attention.
  • Chat transcript — The full conversation between the user and the agent.
  • Metadata — If additional metadata was captured (e.g., handoff data), it is displayed in a syntax-highlighted JSON view with keys visually distinguished for easy reading.
Use the conversation panel to quickly audit individual sessions without navigating away from the table.

Exporting Data

You can export the filtered conversation data for offline analysis or reporting.