Summary Type Score
IAM roles, RBAC, tokens provisioning can take 3-times the effort vs prompting for assessment. Is there an enterprise approved IDE-AI addon (Cursor, windsurf, co-pilot)? Concern 15
Severe level of unknown infrastructure can delay provisional contract with a consulting company for implementation Risk 22
Do we trust the 'private' claim from Microsoft - this is OpenAI in the first place? Concern 15
Lack of data catalogue tool can double the effort. Do we need Atlan first, in parallel the worst? Concern 20
Security and privacy - design and governance ownership - who is signing off initial embedding of OpenAI on each enterprise data endpoint? A thorough level of architecture needed ahead of executive sign off! Risk 25
Do we need a consulting 3rd party at all, where is the value actually? Concern 12
A heavy exercise to do the initial setup - what do we do post assessment phase? Can we leverage Private AI for more value here - egg. plug and play schemas for ingesting separated data? Concern 10