Overview

Conversational Search is the chat interface in the Pieces Desktop App. You ask questions about past work, and responses draw on context captured by the Long-Term Memory (LTM-2.7) Engine.

You can include all captured memories or limit scope by app, time range, and modality. You can also attach local files and folders. Chats are saved to Pieces Timeline so you can open them again later.

Home view with Conversational Search, suggested chats, recent chats, and Start New Chat

Conversational Search homepage with suggested chats, recent chats, and Start New Chat

Getting Started on the Homepage

When you open Conversational Search from the Desktop App home view, you have three entry points:

  • Suggested Chats for You — Conversation starters based on recent activity. Click one to open a thread; responses include Related Timeline Events showing which memories were used.
  • Resume Recent Chats — Cards for your latest threads. Click any card to continue with full history intact.
  • Start New Chat — Type in the input at the bottom and press Enter to ask your own question.
Click `Refresh suggestions` (circular icon beside the suggestions grid) to regenerate prompts from your latest activity. Click `Reveal previous chats in timeline` (panel icon on the *Resume Recent Chats* row) to open older threads in [Pieces Timeline](/products/desktop/timeline).

Example questions:

  • "Why did we choose PostgreSQL over MySQL for the auth project?"
  • "What was the blocker I hit last Tuesday afternoon?"
  • "What did Sarah and I discuss about the API redesign in Teams?"

While You Chat

The bottom toolbar controls how each conversation runs:

Control What it does
+ Attach files or folders (or drag and drop into the input)
Filter By... Limit memories by app, time, or modality
Model selector Pick a cloud model mode (Fast, Balanced, or Extra Thinking)
lotus icon Toggle Reflection Mode for deeper reasoning on hard questions
Send Send a message (Enter queues a follow-up while the agent is still responding)

On active threads, the ⋮ menu at the top lets you Pin, Refresh, or Delete the chat. Token usage appears in the chat header so you can see input, output, reasoning, and cache totals for metered or BYOK plans.

To scope a chat to one Timeline Event or summary, open that item in Timeline and choose Chat from the three-dots menu (⋮). See Add Context to Your Chats.

Explore Conversational Search

Use the Filter By menu to narrow which memories the agent can use in a thread. Attach files, start chats from Timeline Events, and review source memories in the Relevant Summaries sidebar. Switch between Fast, Balanced, and Extra Thinking modes for Claude, Gemini, and GPT. Use keywords, time ranges, app names, and follow-ups to get sharper answers.

If Conversational Search is not what you need, explore Pieces Timeline to browse captured events and summaries, or Single-Click Summaries for one-click workflow reports.