Right now all open AI interaction go through YoBulk open AI keys.In future, we are planning it to be exposed through an ENV variable to make it configurable.

Will it be possible to carry forward context? Like if patterns in my CSV files are persistent, my corrections will be repetitive in nature & can be learnt along.

Looks to be a good tool overall! Where can I find roadmap?

Yes definitely, we are building a Reinforcement learning from Human Feedback based model to save the context based on user's input.So the model will keep on learning form user's correction.

YoBulk roadmap is available here:https://github.com/orgs/yobulkdev/projects/2/views/1 please have a look