How to integrate instructor and tensorzero

[ghbacct]

Is your feature request related to a problem? Please describe.
Setting up additional ML Ops infrastructure around structured extraction workflows can be cumbersome, e.g. recording input output pairs, tracking user feedback, building a model-improvement flywheel using this data.

Describe the solution you'd like
I'd like to use instructor's awesome structured extraction functionality while also benefitting from tensorzero's easy to set-up infrastructure for tracking LLM inferences, episodes and user feedback .

Describe alternatives you've considered
Manually setting up custom logging, feedback collection and model re-training workflows and APIs.

Additional context
This is an open question that I'm currently thinking about and would love to receive some feedback from the wider community.

[ghbacct]

Related issue: tensorzero/tensorzero#549