​
Workflows

Workflows are the building blocks of Itzam. They represent a specific functionality in your app (e.g. a support chat, a code generator, a content writer, etc). A Workflow contains:
  • Model: The AI model you want to use (e.g. GPT-4o, Claude 3.7 Sonnet, Gemini 2.5 Pro, etc).
  • Prompt: The prompt you want to give to the model (e.g. “You are a customer support agent from Acme Inc…”).
  • Knowledge: The files and links you want the AI to use as context (e.g. company docs, wikipedia links, etc).
  • Runs: The runs are the executions of a workflow.

​
Runs

Runs are the executions of a workflow. You can think of them as the logs or history of a workflow. A Run contains:
  • Input: The input (i.e. question) your user provided to the model.
  • Output: The output (i.e. answer) the AI model generated.
  • Prompt: The prompt you provided to the model.
  • Knowledge: The files and links the model used as context to answer the question.
  • Cost: The cost of the run (in dollars).
  • Duration: The duration of the run (in milliseconds).
  • Metadata: Other metadata about the run (e.g. input tokens, output tokens, etc).

​
Playground

The playground is a tool that allows you to test your workflow with different settings without changing your live workflow.

​
Providers

Providers are the companies that offer AI models. E.g. OpenAI, Anthropic, Google, DeepSeek, etc.

​
Models

Models are the AI models that you can use in Itzam. E.g. GPT-4o, Claude 3.7 Sonnet, Gemini 2.5 Pro, etc.

​
Providers tokens

Providers tokens are the tokens used to interact with the AI model. There are 2 types of providers tokens:
  • Input Tokens: The tokens you send to the AI model (e.g. your prompt, the user’s message, the context).
  • Output Tokens (or Completion Tokens): The tokens the AI model generates (e.g. the model’s response).
Here you can check more details about providers tokens: AI Tokens Explained.

​
Threads

Threads allow you to easily create a conversation inside a Workflow. A thread is a collection of messages between a user and the AI. You can create as many threads as you want inside a Workflow:

​
Knowledge

Knowledge is the data you provide to the AI model to use to answer a question (e.g. documents, links). This data can be global (used in every run) or specific to a context (used only if the context slug is provided or if the thread is linked to a context).

​
Contexts

Contexts are a way to group knowledge. You can use contexts to provide knowledge to the AI only when you want to. For example, you can have a context for “Customer Support” and a context for “Sales”. Then, you can use the “Customer Support” context only if the user is from the Customer Support team You can also link a thread to any context. This will make the context available to the AI model in the whole thread.

​
Prompt

Prompt (or System Prompt) is the instructions you provide to the AI model to use to generate text. E.g. “You are a customer support agent from Acme Inc…”.