Context And Artifacts
Agents work best when their task is grounded in artifacts, links, blobs, and events that people can inspect. Veoveo keeps those objects in the workspace graph so bot output can be traced back to source context.
What Counts As Context
- The current workspace and selected bot.
- The artifact a user is viewing or names in the prompt.
- Artifact metadata, kind-specific data, links, and events.
- Source blobs, rendered views, transcripts, datasets, thumbnails, and tool outputs.
- Relevant operations, messages, prompts, and prior decisions.
How To Ground A Request
Ask for work against a visible artifact, route, operation, or specific question. For example, ask a bot to summarize a document artifact, compare related events, inspect a dataset, or create a link between two artifacts. Avoid broad prompts that require the bot to guess the operating context.
Review The Source
Agent responses should point back to artifacts, tool outputs, or events when possible. Before acting on a result, inspect the source material that supports it.
Media To Add
- Screenshot: bot dock opened beside an artifact detail page with the selected artifact visible. It helps users understand grounded prompts. Source: sample artifact detail page.
- Diagram: prompt context resolving to artifact metadata, blobs, links, and events. It helps developers understand context boundaries. Source: sample operation trace.