Starting a conversation

Open the agent with Ask Alva, or start typing on your start page. An empty conversation offers three starting points - Create a new job position, Create a new interview template and List new active job positions - but you are not limited to them.

+ New conversation starts a fresh one. Conversation history holds your earlier conversations, grouped by Past week and Older, and Copy chat history copies the full exchange as text so you can paste it into a document or a ticket.

💡Note: Each conversation is yours. Conversation history is per person, so a colleague cannot open a conversation you started - use Copy chat history if you need to hand one over.

Writing a good request

While you type:

/ lists the skills available to you, marked Global or Organisation depending on where they come from.
@ mentions a job position, a candidate, a to-do or an employee, so the agent works on the right one.
Attach files adds a document to the conversation. You can also drop files straight onto it.

The agent also picks up the page you are on, so you can ask about the job or candidate in front of you without naming it. A chip above the message box shows what it is using, and you can remove it if your question is about something else.

💡Note: Skills are what make the agent follow your procedure rather than a generic one. See Agent skills and personalisation.

The cards the agent produces

When the agent produces something substantial it appears as a card in the conversation - a Job advert, Job description, Skills & experiences, Assessment plan, Screening questions, Interview or Readiness screening.

The card is a preview, not the real thing: the full, editable version lives in the job itself, and the card tells you so. Show artefacts brings the panel back if you have collapsed it.

Deciding what the agent may do on its own

Permission mode controls how far the agent goes before it checks with you:

Require approval - the agent always asks before it takes an action on your behalf.
Run automatically - the agent handles the work on its own and only asks you if something looks risky.

You set it per conversation from the agent panel, and separately on each workflow.

When the agent asks, you get a Tool approval request with two answers: Yes, I approve, or No, and tell Alva what to do differently - the second one lets you write what should happen instead. View details shows exactly what it wants to do before you decide.

💡Note: Requests raised by a workflow do not appear in a chat, because nobody is sitting in one when a schedule fires. They wait in the Action hub - see Agent workflows and the Action hub.

Moving the chat around

The panel does not have to stay where it opens. At the top of the Agent chat, when docked to sidebar, you can move the chat around:

Pop out to floating window turns it into a window you can drag and resize.
Dock to sidebar puts it back.
Open full page gives it the whole screen.

When a conversation runs out of room

A conversation has a limit. When it is reached you will see This conversation is full; click Start a new conversation to keep going. Anything the agent already created stays where it was saved.

You can also interrupt the agent at any point: Stop ends the current turn without closing the conversation.

For what the agent is and where to find it, see An Introduction to Alva Hiring Agent.

Any Questions?

Use the chat in the bottom right corner to connect with a member of our support team.

Chatting with Alva Hiring Agent