Hi Angela, great questions! Happy to shed some light on how Chat for Blackbaud AI works.

On feedback and how quickly it's incorporated: The thumbs up/thumbs down feedback you're providing is reviewed by our team (with sensitive details redacted as needed) and used to improve the AI chat tool over time. It's not incorporated in real time, but it does directly inform ongoing improvements to the product.

On training your instance vs. network-wide learning: You don't need to train your own instance. Importantly, Blackbaud AI does not train on your organization's data — your prompts and responses stay within your environment. The AI generates responses based on the context of your Raiser's Edge NXT data while respecting your organization's existing security permissions.

On surfacing contextual recommendations like fund-to-donor affinity: This is a great use case you're describing — and it's actually an area we're actively working on! The ability to connect things like notes on a fund's intended purpose with a constituent's strategy notes to surface relevant gift opportunities isn't fully available just yet. Keep an eye on the Prompt Library, though — when these new capabilities are ready, we'll be adding example prompts there so you'll know exactly how to take advantage of them.

Tip: The more you can surface details within first class fields like Philanthropic Interests rather than buried in notes is going to help AI connect more dots.

Thanks for testing and sharing your feedback — it's exactly the kind of insight that helps us build the right things!