iostack.ai Portal

iostack Agents are built by constructing a graphical model of your conversation process.

When you first create an Agent, it will be pre-populated with a starting Stage and that stage will be configured as the entry point for the Agent i.e. this is the first stage that will be used to control interactions between the LLM and the user.
It is extremely easy to then extend your Agent by adding new stages and transitions using drag-and-drop.

Agent-wide and common stage settings can be configured via the Settings and Common Stage Settings sections respectively.
For more details on these sections, refer to the Settings and the Common Stage Settings sections.

Stage and Transition settings are accessible by either clicking on or moving your mouse over the entity in question and clicking on the pencil icon.
For more details on Stage and Transition settings, refer to the Stage and Transition Setting sections. section.

The Portal displays the current Saved status of your agent at the top right of the Portal, adjacent to the Sandbox.
Saving changes to the Agent will happen automatically when the Portal detects that you have not been making changes for a short period.
To manually trigger saving the Agent, either click on the Saved status icon or type CTRL-S.

To test your Agent, click on the play button in the Agent’s Sandbox.

The Agent will be compiled and run. If any errors are encountered during this process due to missing or incompatible settings, they will be reported and can be browsed and fixed.
You can interact with your Agent using the message box to send it messages and view the responses in the transcript of the conversation.
You will also be able to follow where in your conversation process the agent is, as highlighted on the flow diagram itself.
For full details on using the Sandbox to test and debug your agent, refer to Sandbox documentation.

Building, Editing & Debugging