This release makes the agent faster to reach and more capable with your source material. You can now pop chat into a floating window, drive the whole app from a ⌘K command palette, and attach large PDFs without them silently going nowhere.
Floating Chat + ⌘K Command Palette
The agent no longer lives in a fixed side panel. Pop it into a floating window (movable, overlay-style, Notion/Linear style) and keep it visible wherever you're working in the app. Especially useful on laptops, where the side panel always felt a bit cramped.
Alongside the floating chat, a new ⌘K command palette surfaces contextual actions for whatever page you're on. Each action has an optional keyboard shortcut, so common tasks like sending a test broadcast, scheduling a send, or launching a campaign are a keystroke away instead of a menu hunt.
Large Files in Chat
Drop in a large PDF or document and the agent will actually respond. Previously, big attachments would upload fine but the agent would silently stall: the file exceeded the provider's request-body limit before it ever reached the model.
Files are now delivered by reference at send time rather than inlined, which clears the limit cleanly across Claude, Gemini, and OpenAI. Drop in a report, a deck, or a brief and ask the agent to summarize or pull from it.
Other Improvements
Agent: Bulk-update task statuses, bulk-update people from a segment, batch-create X reply threads, and search MCP tools by keyword. Big multi-item jobs in one go. Agent questions now support multiple-choice and free-form answers together.
Audience: People and Companies tables now sort and paginate server-side, snappy even for very large audiences. Company pages load faster for companies with lots of people and deals.
Chat: Search your chat history in a new popover, grouped by Today / Yesterday / Previous 7 & 30 Days. Send attachment-only messages (no text required). Cleaner scheduled-social-post confirmations, including a LinkedIn first-comment preview.
Developer Settings: Redesigned API keys page: one clean table (name, type, key, last used) with details and key creation moved into side sheets, replacing the old sprawling dual-table layout.
Memory: Rename documents and folders while editing, without losing your place.
Social: X threads retry safely without creating duplicate posts after a flaky network error. More accurate LinkedIn reconnect prompts (no more false "reconnect your account" nags).
Plans: A real trial plan for new signups, with a clearer trial lifecycle.
Blog
Keep up with the latest
from Quotient
Similar Topics