Talvi Recipes
Practical workflows for building, updating, cleaning, and extending your AI agents.
Choose a recipe
Follow these recipes to complete a task across Talvi’s administration screens. Each recipe explains the order of work, what to check, and how to recognise the result.
| I want to… | Recipe |
|---|---|
| Give a new agent a knowledge base | Build a dataset and create an AI agent |
| Add or refresh knowledge | Update your AI agent’s data |
| Remove a source from the system | Unpublish and remove content |
| Review similar files regularly | Clean duplicate and similar files |
| Give an agent a third-party capability | Add an MCP tool to an AI agent |
| Add a chatbot to my website | Add your Talvi AI bot to your website |
| Give people access to an agent | Assign users to an AI agent |
| Share useful learning content | Assign and distribute Talvi Nuggets |
| Turn Nugget activity into improvements | Create a Talvi Nugget AI analysis report |
Before you start
A dataset is the knowledge base for an AI agent. One dataset can be connected to many agents. Check the linked agents before importing, unpublishing, or deleting files: a change to shared knowledge can affect them all.
The screens available depend on your permissions and Talvi deployment. Create Dataset, AI Agent Studio, and MCP Tool Store are in the super admin workspace when enabled. Ask your administrator if a required control is missing.