What is Workers
Workers functions as a personal workspace with persistent conversation sessions. Each session is linked to a specific agent or squad in your workspace. The agent responds using the same LLM model, tools, and knowledge base it uses in real customer service — the difference is that the conversation is yours, private, and does not generate records in the customer CRM.When to use Workers
- Data and conversation analysis: ask the agent to process a large volume of records, extract patterns, or consolidate information into a structured report.
- Research and information synthesis: ask the agent to research a topic, compare sources, and deliver a ready-to-use summary.
- Batch content generation: delegate the creation of multiple texts, scripts, templates, or replies all at once.
- Tasks with external integrations: when the agent needs to access Google Sheets, Notion, Slack, or other tools, processing may take longer — Workers ensures the task is not interrupted.
How it works
The Workers screen is divided into two main panels that you will use together during work.Left panel (conversation history)
The left sidebar lists all your sessions in chronological order, with the most recent at the top. Each item displays:- Name of the agent or squad executor
- Preview of the last message in the session
- Time of the last activity
Main chat area
The central area displays the active conversation with the selected agent. The message field accepts text and attachments (images, PDFs, Office documents, and CSVs) with a limit of 25 MB per file and 4,000 characters per message.Starting a new conversation
Click the new session icon
At the top of the sidebar, click the + button to open the new session dialog.
Choose the executor
The dialog displays two tabs: Agents and Squads. Select the agent or squad that will process the task. Each option shows the name and avatar of the executor.
The Credits balance available for the workspace appears in the footer of the message field. Each processed message consumes Credits according to the LLM model configured on the agent.
Continuing conversations
Previous sessions stay in the sidebar and can be resumed at any time:- Click a session to open the full history and continue from where you left off.
- The conversation context is preserved between sessions — the agent “remembers” what was said previously.
Renaming conversations
To facilitate organization, you can rename any session:Search
The search field at the top of the sidebar filters sessions by content in real time:- The search considers the session title and message content.
- Results appear instantly as you type.
- To clear the filter, delete the text in the field or click the close icon.
Behavior in production
Workers runs the same agents that are active in real customer service. This has important implications:- Messages sent in Workers consume real Credits from the workspace.
- Files sent are stored in the workspace’s storage bucket.
- External integrations triggered by the agent (such as Google Calendar or Sheets) execute real actions — not in a test environment.
Workers vs. Internal Chat
| Feature | Workers | Internal Chat (Time AI) |
|---|---|---|
| Executor | Agent or squad from your account | Timely.ai built-in copilot |
| History | Separate sessions per executor | Unified sessions |
| Consumes plan Credits | Yes | No |
Best practices
- Write the first message with complete instructions to reduce back-and-forth with the agent.
- Send all relevant files in a single message when possible — the agent processes the set as unified context.
- Create separate sessions for distinct tasks. Mixing topics in a session makes future reference harder and can confuse the agent’s context.
- Use Workers for tasks that take more than a few minutes. For quick queries, the chat in the agent panel is more agile.
Conclusion
Workers is the shortcut to leveraging the power of your agents beyond customer service:- Personal productivity with autonomous processing of time-consuming tasks.
- History organized by session, agent, and period.
- Support for files and multimodal vision when the agent’s model supports it.
- Context preserved between sessions for ongoing work.