track_context()

track_context() creates a context manager which defines a Pay-i instrumentation scope. Every inference request in this scope will be instrumented by Pay-i with the properties defined.

Parameters

All parameters are optional and default to None

Parameter	Type	Notes
`limit_ids`	`list[str]`
`use_case_name`	`str`
`use_case_id`	`str`	If `None`and the passed `use_case_name` 1. creates a new payi instrumentation scope 2. or is different value comapred to the current payi instrumentation scope Then in a new `use_case_id` will be automatically generated for the new payi instrumenation scope. This evaluation occurs every time the decorated function (`do_work()` in the example) is called.
`use_case_step`	`str`
`user_id`	`str`
`account_name`	`str`
`request_properties`	`dict[str, str]`
`use_case_properties`	`dict[str, str]`
`price_as_category`	`str`	Override the payi Category for the instrumented inference calls. Useful when using the `OpenAI` client to send requests to non `OpenAI` models.
`price_as_resource`	`str`	Override the payi Resource for the instrumented inference calls. Useful when the inference calls specify a deployment name, not a model name.
`resource_scope`	`str`	Applies a deployment scope to the instrumented Resource . Example values (but not limited to) are `global`, `datazone`, and `region`.
`log_prompt_and_response`	`bool`	Overrides the global `log_prompt_and_response` setting created by payi_instrument()

Example

pip install payi openai dotenv

from payi.lib.instrument import track_context, payi_instrument
from openai import OpenAI
from dotenv import load_dotenv

load_dotenv()

payi_instrument()

openai = OpenAI()

# A new use_case_id will be generated
with track_context(use_case_name="teach_me", use_case_step="one"):
  response = openai.chat.completions.create(
    model="gpt-5.5",
    messages=[
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "Hello, how can I use the API?"}
    ])

  # Same use_case_id will be used since use_case_name has not changed
	with track_context(use_case_step="two"):
  	response = openai.chat.completions.create(
    	model="gpt-5.5",
    	messages=[
      	{"role": "system", "content": "You are a helpful assistant."},
      	{"role": "user", "content": "Tell me a one line joke"}
    	])