I'm trying to test out the AI Builder for Form Processing. I've tried simple PDF's and simle PNG scans of the same form. For each model, after clicking Analyze, I get a message "Couldn't analyze documents. Try again later." I've tried it over several days, but the model never trains. Has anybody gotten it to work?
Solved! Go to Solution.
Thank you Topness,
Your video helped me figure out what's going on. You go above the limit of 4MB per data set you provide. Please try with only the first 5 image documents and it should work fine.
In the meantime, we're improving our wording inside the model creation flow to be more explicit about the limitations which are for the whole dataset:
And thank you for the hint about fillable PDF, I'll forward the information.
I'm getting a status error 404 for https://graph.windows.net/me/thumbnailPhoto?api-version=1.6 and also: "Failed to load resource: net::ERR_NETWORK_CHANGED" for https://browser.pipe.aria.microsoft.com/Collector/3.0/?qsp=true&content-type=application%2Fbond-comp...
Are you getting the same errors? Maybe the problems are there, anyways I don't know how to fix this.
I'm also getting GET https://graph.windows.net/me/thumbnailPhoto?api-version=1.6 404 (Not Found) when the model loads. I'm also getting this when it loads:
Refused to execute script from 'https://make.powerapps.com/e/Default-ea7f704e-aaa4-4dd8-9bec-8654a814fc02/aibuilder/models/<GUID>/pd...' because its MIME type ('text/html') is not executable, and strict MIME type checking is enabled.
Which might actually be a 404 too.
The error isn't explicit and the AI Builder team is improving that soon.
In the meantime, please check that the input documents you provide respect the current limitations of the service as per the documentation.
I'm getting the same error. I've tried in multiple browsers and on multiple days, to make sure it wasn't a temporary glitch. I've also verified my PDF sizes total less than 1mb for all sample PDF's combined, per the MS documentation. And I've verified I have the proper access rights in D365 to Core Records in the Note entity.
Draft models that failed analysis aren't being correctly recovered. It is recommended to delete those drafts and start building another model from scratch.
Hi Mustlaz. I've spent some more time trying to get this to work. My company has a Prod, Test and Sandbox environment and I've tried in all 3. Always, I get the same error. I've tried with different PDF documents. And I've tried deleting my existing failed AI models and starting from scratch. Always the same error.
Same here. I'm in a trial tenant, I've tried the default env and a new environment. I've reloaded my documents (a simple form scanned to PNG and several filled PDF's). Both continue to return "Couldn't analyze documents. Try again later." after analyzing in a new model. For reference, the files I'm using are here on Github.