Skip to main content

Chat vs. Lab vs. Actions

Chat uses the agent’s active version — the same configuration teammates get when they pick this agent for production work. It is not a generic assistant; it is your specialist with your prompt, Focuses, and linked context.

Open Chat

From the Agent Manager hub, choose Chat. Select the agent in the top left. The conversation uses that agent’s active configuration.
Chat page

Before you send

What happens during a conversation

Behind the scenes, the agent runs the observe → think → act loop from What agents are — and aren’t:
  1. Your message joins the prompt stack (system prompt, linked files, history, tools).
  2. The agent decides whether to reply, call a tool, or create an artifact.
  3. Tool results and artifact updates feed back into the next turn.
You may see:
  • Streaming replies as the agent composes an answer
  • Tool activity when it looks up data, runs an Action, or opens a builder
  • Artifacts in a side panel when a builder Focus creates a document, table, or other output
  • Reasoning/thought blocks on supported models — collapsible summaries of intermediate thinking
Long conversations fill the context budget. If quality drops, start a new chat or trim linked context on the agent configuration.

Attach files in Chat

Use the attach control on the composer to add chat-linked files for this thread only — contracts, spreadsheets, images, and other supported types. The agent can reference them for the rest of that conversation. Permanent reference material belongs in linked context on the agent, not re-uploaded every chat.
Video placeholder Short screen recording: select agent → attach a PDF → ask a question → agent cites content from the file. 30–45 seconds, no narration required if captioned.

Realistic expectations

Agents are drivers, not rails. The same question can produce different phrasing or route choices. That is normal. When correctness matters, pair Chat with human review, deterministic Actions, or Track approval milestones.

Where to go next

Chat Review

Revisit and audit past conversations.

Working with artifacts

Outputs saved from agent conversations.
If you would like something unique to your business or have a unique issue, use XSpada Evolve or ask a XSpada rep.