What the assistant can do
Behind a single question, the assistant can:- Search your knowledge base — the documents and data your team has uploaded to Knowledge.
- Search the web, with the option to scope results to specific pages or turn web search off entirely (see Sources and attachments).
- Query market intelligence and social media data.
- Query your portfolio database, if your organization has Portfolio brand data.
- Recall past conversations and sources it has cited before.
- Analyze data and create files — spreadsheets, slide decks, documents, and interactive visualizations (see Generated files).
- Ask clarifying questions — an inline form appears in the chat with Yes/No buttons or text fields. Answer the ones that help, or skip them.
- Propose Trackers — when a question suggests ongoing monitoring, the assistant can offer an inline proposal you can accept to create a Tracker.
- Run deep research — in Research mode, the assistant produces a full formatted report with a generated cover illustration (see Modes and outputs).
Every chat screen carries the reminder that Merciv is AI and can make mistakes. Double-check cited sources before acting on an answer.
Starting a chat
You can start a conversation from several places:- Click the New chat plus button next to Recent Chats in the left navigation.
- Type into the composer on the home page — this creates a new chat automatically.
- Open the chat as a side panel over any page in the app. Use Expand to Full Screen and Move to Side Panel to switch between the panel and the full-screen view.
- A brand-new chat shows a carousel of prompt templates you can click to get started.
- Your chat is named automatically based on the conversation. You can rename it any time.
- Messages are limited to 100,000 characters.
- While the assistant is working, the send button becomes a stop control — click it to interrupt the response.
- Longer tasks show rotating status phrases. Deep analysis can take a few minutes, so “Still working through this one…” after 30 seconds or so is normal.
The activity timeline
Each response includes an expandable Activity section showing the assistant’s thinking and each action it took — searches it ran, files it read, code it executed. When the response finishes, the section reads Activity · N actions. Sticky indicators show running counts of sources, files, and to-dos as the assistant works. If something goes wrong mid-task, an inline Task failed indicator appears; sending the request again usually resolves it.Citations
Answers include numbered citations, and a sources strip at the bottom of each message lists everything the assistant cited. When you copy an answer with Copy Text, the sources come with it.Chat history
- Your chats appear under Recent Chats in the left navigation, sorted by most recent, and in project and home lists.
- Within a long chat, older messages load 10 at a time — click Load more at the top of the conversation.
- A chat can belong to a Project. Chats started from a project page attach to that project automatically.
Renaming, moving, and deleting chats
Open the ”…” menu on any chat to:- Open the chat
- Rename it
- Move to Project — file it into an existing or new project
- Delete it — the chat disappears from your lists, and its link no longer works