Hey ​@vamsi 

We had the same problem. Our solution was to group the sticky note with the shape. As groups are exposed on the API you can re-construct the ‘connection’.

To ensure all notes are grouped, we introduced a conditional formatting rule to spot those.. 

Hope this helps - Cheers Tobi

Hi ​@tobi,

Thanks so much for the suggestion! Using groups to create an explicit link is a fantastic idea, and the conditional formatting rule to enforce it is a great process tip.

I ran a diagnostic script against our target document to check for any "Group" objects, and it turns out the diagrams we need to process don't use this grouping convention. Unfortunately, since we don't have control over how these source documents are created, we can't enforce this as a standard.

Do we have any other approaches which can cleverly identify the sticky notes which is near to any shape, and somehow that can be appended to the flow.

hey ​@vamsi - ah right, well I assumed you have control over the source :( Well in this case I can’t help further. I’d need access to the x/y coordinates of the shape which ASAFIK are not exposed on API.

but wait, doesn’t the API allow export as PDF/PNG export? One (dirty/one-time) approach could be to feed that to LLM and extract the connections?

You're right, the export option is the next logical step. My only worry is that our flowcharts can get quite large and complex. I'm worried that when we save them as an image, some of the smaller text might become illegible or the AI might get confused by crowded connections, leading to an inaccurate analysis.