Aaron J.

I need a human agent to check. I submitted all assignments before the deadline.

Izzy FYI this might be an interesting case study for future cohorts

Max M. this touches on a workflow we are exploring in the AI cohort. You can pass latitude and longitude into the following URL format (https://www.google.com/maps/@(LATITUDE),(LONGITUDE),200m/data=!3m1!1e3) where 200m indicates a viewspace of 200 meters. Then you can use 'capture screenshot' enrichment based on that url to get a screenshot of google maps. Then use 'analyze image' enrichment with specific instructions to identify solar panels in the image (estimate sq meters maybe?). You may need to pass that output into an AI text model (GPT 4.1 mini for example) to parse out specific results. I would recommend you find an example or two where solar panels are visible on google maps, and use those as a test case to refine your prompts until its giving you consistent accurate results.

Aladin the server seems to max out at 500 title results total, so I was only able to pull the first 500 for you. I wasn't able to find a method that returned the full list. I would estimate there are 3000+ total titles in the underlying directory.

Tim P. I'd agree with the above post, but a lot of it comes down to personal preference and testing. For example, I found Gemini to be better at parsing call transcripts for compliance purposes than Claude or ChatGPT, and I prefer Claude's writing style over Gemini or ChatGPT. Also, since the space keeps expanding and changing, its hard to predict which new models will be better for certain tasks. I recommend coming up with a testable use-case (you already know the correct answer) and test different models against it using the same prompt. This will give you a better feel for what works best for certain tasks.

No, because it would give them admin-level access to systems we do not currently allow them to access as admins (salesforce for example)

No we are not on an enterprise level plan.