Attach Agents and Calls
Instead of describing which agent or call you mean, attach it. Studio reads its real data and answers about that.
How to attach
Click the + button next to the message input, or type @ in the message box. Search by name and pick what you want. Each attached item shows as a chip above the input — click the × on a chip to remove it.
You can attach:
| Type | How many | What Studio sees |
|---|---|---|
| Agents | Up to 5 | The agent’s configuration, prompt, welcome message, attached knowledge, and tools |
| Calls | Up to 10 | The call’s outcome, transcript, and metadata |
Why it helps
Once something is attached, you can ask about it directly:
“Why did this call fail?” “Tighten this prompt.” “What’s different between these two agents?” “Did the agent collect the phone number on any of these calls?”
Studio treats the attached items as the subject of your message, in the order you attached them.
Studio fetches attached items at the moment you send the message, so you’re never looking at a stale copy. It also won’t waste a step looking up something you’ve already attached.
Useful patterns
Debug a bad call. Attach the call and ask what went wrong. Attach the agent too, and Studio can tell you which part of the configuration caused it.
Compare agents. Attach two and ask what differs — faster than opening both and eyeballing the settings.
Review a batch. Attach several calls from the same campaign and ask what callers kept asking about, or where the agent kept getting stuck.
Improve a prompt with evidence. Attach the agent plus three calls where it went wrong, and ask Studio to suggest prompt changes grounded in what actually happened.
What happens if something is deleted
If you attach an item and it no longer exists by the time you send, Studio tells you it’s unavailable. It won’t quietly substitute a different agent or call.
Attached context vs. product questions
Attaching tells Studio what you’re asking about — it doesn’t change how Studio answers questions about Dialora itself. Product answers are still grounded in the official documentation. So “why did this call fail?” uses your attached call, while “how does voicemail detection work?” is answered from the docs, whatever you have attached.