Chat

Overview

Chat is your entry-point into working with the Attive agent inside a workspace. It’s where you ask questions, get answers, and iterate—while keeping the conversation grounded in the context you choose (like agents, memories, insights, and documents).

Usage

General usage

• Start a chat: Open Chat and type your question.
• Add context: Use the Memory sidebar and/or select an agent before sending a message.
• Attach files (optional): Add relevant documents to help ground the response.
• Iterate: Ask follow-ups, collapse older answers, and keep refining your question.
• Revisit: Use Chat History to return to a previous conversation.
• Share: Share a chat to make it easy for others to review (read-only access may apply).

Document folders (how to add them, and how they work)

Document folders let you give the agent access to a curated set of documents—so it can search within them and pull relevant excerpts into its answer.

• How to add document folders

  • In the chat composer, open the Folders menu (the folder picker in the composer toolbar).
  • Choose one or more document folders you want the agent to use as reference.
  • Your selected folders appear as chips in the composer, and you can remove them at any time.
  • Some folders may be marked as always available—those are automatically included as default context.
  • Folder selections typically persist as you continue the chat, so follow-up questions stay grounded in the same source set.

• How folders work from the agent’s perspective

  • The agent receives a list of the available folders (and the files inside them), and may also receive pre-selected relevant excerpts based on your current question.
  • The agent can use document tools to:
    • List what’s in the selected folders
    • Search within the selected folders
    • Read specific chunks of documents for quoting and verification
  • Importantly, folder access is restricted to the folders you’ve selected for the chat—so the agent won’t browse outside that scope unless you add more folders.

• Folders vs file attachments

  • Folders are a scoped library the agent can search across.
  • Attachments are specific documents you include with a message (useful when you want the agent to focus on a particular file right now).

Memories (default mode vs when an agent is selected)

Memories are short pieces of context (facts, preferences, definitions, important notes) that help Chat stay grounded and avoid repeating work.

• Default mode (no agent selected)

  • Chat maintains a conversation memory context that can grow as you ask questions.
  • By default, the context is auto-updating: Chat can automatically pull in relevant memories for your current question and add them to the conversation context for future turns.
  • You can lock the conversation context to stop it from changing automatically.
  • You can also add memories explicitly (for example, from the “All memories” list) when you want to ensure a specific memory is included.

• When an agent is selected

  • The agent’s pre-configured memories become the base context for the conversation.
  • Automatic memory retrieval is disabled (so the chat doesn’t keep expanding context on its own).
  • The Memory sidebar becomes read-only for agent context, and you’ll see an Agent tab showing what that agent is configured to use.

• How to use the Memory sidebar

  • Chat: what’s currently active for this conversation (available in default mode).
  • Matches: what memories were recalled for a specific message.
  • Suggested: memories the system proposes you save/enable.
  • All: browse all workspace memories and add them to the conversation context (default mode).

Examples

• Exploratory question

  • “What changed in pipeline health this quarter, and what are the top drivers?”

• Context-driven analysis

  • Select an agent focused on GTM analysis, add a few relevant memories/insights, then ask:
  • “Summarize the key risks for renewal accounts and suggest next actions.”

• Document-grounded request

  • Attach a document and ask:
  • “Extract the key takeaways and propose an outreach plan based on this content.”

Features

Feature	What it does
Threaded conversations	Organizes messages into question + answer pairs, with controls to collapse/expand answers for easier scanning.
Real-time responses	Streams answers as they’re generated, supports resuming interrupted responses, and lets you stop mid-stream.
Memory & context controls	Lets you view/manage applied context and keep it consistent (for example, by locking context).
Agents	Lets you choose an agent to guide how the assistant responds based on agent-specific instructions and context.
Chat history	Lets you revisit and jump between previous chats in the workspace.
Sharing & read-only access	Lets you share a chat for others to view; shared chats may be read-only depending on access.
File attachments	Lets you attach files to ground responses in specific documents and content.
Retry on failure	Lets you retry sending a message or regenerate an answer when something fails.

Related Features

• Agents: Configure or select an agent to shape responses and apply agent-specific context. See Agents.
• Memory: Curate the context the assistant should use (and optionally lock it). See Memories.
• Prompts: Reusable starting points for common questions. See Prompts.
• Boards & Insights: Save and collaborate on results from Chat. See Boards and Insights.