Roadmap

Want to be able to add myself to the client team on eu.nanonets, eu-open, in.nanonets. Currently have to ask someone from tech to add us.

Customer wants to use google translate chrome plug-in to translate Chinese language documents. This will enable them to read and sense check / correct annotations. Adding call recording below for more details.

Unable to rename table headers with spaces in whitelabel

https://www.loom.com/share/f44b11f32bc84f29bd10609aaad4408e?sid=1488ef2c-f95b-4b48-a9a9-f506d2f22d80 I have a data action (call it DA1) using nanonets ai as input. When I add a DA above this (call this DA0), for same field, the input for DA1 does not automatically change to the output of DA0. Both DA1 and DA0 use nanonets ai label id for input.

Santos estate has 'advanced edit access' enabled. Most of their regular users don't have access to edit 2-3 fields reserved for their finance approvers. We have some calculations etc in Python PP; and for many files, after making corrections, the users will have to rerun workflow. However, we block these rerun workflow calls if the users don't have access to any field in moderated boxes. Currently on assigning file or making any edit, moderated boxes get populated with all the data. Subsequently users are unable to rerun workflow

User added lookup from CSV with AI match. Then they uploaded some files and approved. This populated the vector_db with entries from those files. Next, user added a test file. Here, the lookup input value was different from the previously approved files (therefore not present in vector db). But exact value was present in CSV. So user expected it to match. But it ended up matching with one of the existing rows from vector db. Giving an incorrect match when EXACT input value is present in CSV is a terrible UX

Issue in Data actions, export and possibly PP(?) as well. The input shown in the sample file selection at the bottom of Python script is not the same as the input actually used in the code. This makes it difficult for any external user to actually write script by themselves, and there is no indication that the input is wrong.

No way to create a oneshot model by copying training files. Have to download training files and annotations and from s3 and reupload in a new model. Since we are actively using and editing training files now, makes sense to create a new model using them.

Hi Sravanth, yesterday I was retraining the model on a PO. As I was scrolling through the table data, my cursor inadvertently moved over the "Train Model" icon just below the table data scroll bar. It immediately went into training mode. I didn't want to train the model because it meant I wouldn't be able to train it at the end of the day because of the amount of time it takes to train. Is there a way to stop a training session in such cases? It doesn't give you a warning message when you click on "Train Model". It would be really great if we could select the one customer to train because sometimes no changes are made other than to just one customer.

Transform how you interact with data through our Natural Language Analytics. Simply type your query, and our system will not only visualize the data but also generate comprehensive reports on the fly. Make data-driven decisions effortlessly and with precision.