"Building an AI-Ready Content Library: Your Blueprint for Success" – Coffee Chat Recap

Helene H

Thank you to everyone who joined today’s Coffee Chat! With our highest attendance to date, it's clear that Content Management + AI is a topic you’re all excited about.

The recording and resources are now available in the Academy, and the questions are in the discussion thread!

Key Takeaways:

• Efficient AI tools depend on an organized library
• Clear, distinct content reduces misunderstandings
• Consistent structure improves AI performance

Special shoutout to @Nicole Stanfordfor sharing her PS expertise!

🚀 Congratulations to our 5 Certification Winners 🎉

Let us know what other topics you'd love to see—don’t worry, prompting tips are already on the list!

Up Next: Curious about Responsive’s vision for AI in SRM and what’s on the roadmap? Join us on April 23 for our next Coffee Chat: "The Future of AI in Strategic Response Management" featuring:

• Kunal Shrestha, VP of Product at Responsive
• Jennifer Cannizzaro, VP of Product Marketing at Responsive

Register now and get your questions answered + feedback heard! 📢

Helene H

1st Pass of the Q&A - following up with any Product specific ones :)

Q: Does AI consider parent questions when generating content for sub-questions in a project?
A: Good question, Honey! It does in the sense that the tag is under the parent, but it won’t search through the breadth of that parent tag's content. I recommend just searching by the parent tag to capture all subtags.

Q: Why does the AI require so much manual input from us to revise the database? I thought AI was supposed to teach itself and improve.
A: Good question, Mike! AI does learn with you, but it still relies on human touch for ensuring content accuracy and relevance. It can find answers quickly, but bad data will result in bad answers. We view AI as a partner, not a replacement!

Q: I am a representative of my organization, but not an admin for our Responsive instance. How do I check if we have AI enabled as an end-user?
A: Hi, Dylan! If you don’t see AI Draft in your projects, the Ask button on any page, or the option to use AI Assistant in the library or response boxes, it likely isn’t enabled. You can also check with your CSM.

Q: For alternate phrasing, should we create an entirely new Q&A pairing with alternate phrasing? We run into that a lot.
A: Hi, Laura! You can create alternate questions within the same Q&A pair. This is our recommended approach unless the topics become inherently different.

Q: Where do I go to set up Datasets? I don’t see it in Responsive AI preferences.
A: Hi, Linda! Your CSM/AM can enable this for you, making it available for you to use.

Q: From a domain, what subdomains does it dig through? Is it limited to that site’s sitemap? Or, if there is a URL in a subdomain, does it "click it" and look around there?
A: Hi, Lisa – great question! You need to list the subdomain in Organization Settings for AI to search it. It won’t automatically search within subdomains from a provided domain.

Q: For removing ROT content to optimize content for AI, what is the recommendation on highly used, older Q&As?
A: Old doesn’t always mean bad! If you have high use on older content and it’s still accurate, that shows you’re doing a great job of maintaining it. It’s all about balance!

Q: When looking at the content search terms report, what does "no action" mean under the action performed? Should it be a concern if a lot of search terms have "no action"?
A: It means the searcher did not apply any responses (unless they typed it manually). It's not necessarily concerning, but ideally, your team should be finding what they need and applying it.

Q: How can I create a field to denote when a Q&A was last reviewed? There’s a Last Updated date, but that’s not what I want. Maybe an SME reviewed a Q&A pair and said it’s perfect as is—how do I note when that review happened?
A: You can perform an advanced filter search in the library or use the customize metadata button to add a "Last Reviewed" field on the Q&A card.

Q: So, are you saying we should generate our own Q&A pairs instead of saving the client version of the question and the sales team's responses?
A: A great suggestion would be to set up Moderation and align your Q&A pairs with how your content library is structured.

Q: How often do you recommend auditing, or is that more business-dependent?
A: At least once a year, but it depends on how frequently your information changes and how much help you have for the audit.

Q: Is there training on how to use the Content Search Term report? There’s a lot of information that comes up.
A: We have an article that I’ll link, and we can also jot this down for a future training: Working with the Content Search Terms report

Q: When will y’all be in Nashville, and what is the occasion?
A: We’ll be there for BCP! Marcea and I are speaking and hosting mini coffee chats at the Responsive booth!

Q: Can you move an existing tag to a hierarchical tag (so that it becomes a nested tag under a main tag)?
A: Yes! You can do this in Organization Settings by selecting Edit on the tag, then Nest. Choose the Parent Tag there.

Q: My AI is pulling in content for projects that are from other non-related projects even when I specifically tag for the library to pull from specific projects.
A: Right now, you can only filter AI using tags, collections, and custom fields, not the project name. If you’re frequently searching specific projects, I recommend structuring tags to support this. Your CSM can also guide you on next steps if needed.

Q: Do content reviewers/owners need to be Responsive "Guests" rather than standard users? (For example, Legal reviewing content but never using the platform for regular use.)
A: You can use the Guest feature for this, but it’s more manual. If they need consistent access to content, I’d recommend pulling them in as Team Members.

Q: When do we expect Gen AI on PPTs (lookup)?
A: AI in Lookup is slated to come in Q2.

Q: Can documents be in a private collection in the Docs library?
A: Yes, you can create a private collection for documents.

Q: Does using Hierarchical Tags improve the AI results? I’m unfamiliar with this.
A: H-Tags can be used to filter AI results, and they allow you to create a folder structure for organizing tags.

Q: Is it useful to clean up customer questions to remove things like "The system Shall" or "the ability to"?
A: It’s up to you. The system typically skips filler words (check out the Stop Words list in Organization Settings). The key is whether it’s hard for your team to find info.

Q: Can AI look also at URLs?
A: You can specify up to 5 Domains/Sub-Domains in Organization Settings for AI to search.

Q: Are the datasets used only by Ask or all of AI?
A: Datasets are currently used only by Ask.

Q: Can you explain a little further how datasets work with databases?
A: Datasets allow you to create a pre-set filter made of Collections + Tags, making filtering in Ask much faster.

Q: Can we load a document for AI to look for answers in it?
A: Yes, AI can search documents. We suggest uploading documents into the document library.

Q: How do you create clear and concise questions if they are coming from a customer’s RFP and they may not be?
A: You can use AI tools to clarify the answers. We suggest using Moderation to update the content before saving it back from a project so it aligns with your content library.

Q: What other metadata does the AI use to identify the proper content? Do we need to structure content a specific way to control AI input?
A: The benefit of custom metadata (Tags, Collections, Custom Fields) is that it helps segment where Responsive AI looks in your content library.

Q: Can ASK search in Seismic if you have the integration?
A: At this time, Responsive AI does not search within Seismic.

Q: We have titles for our questions with actual client questions in the alternative questions. Will this still work in AI?
A: The effectiveness of titles with AI depends on their intent. If the titles are simply labels, it might be better to create a custom field. If the titles are meant to assist with keyword searches, there shouldn’t be much of an impact, but you may want to consider reprioritizing the primary question.

Q: Does this have smart search capabilities?
A: We utilize GPT-4o and other AI models that contain smart search capabilities (OpenAI/Microsoft Azure/Bedrock) for processing queries.

Q: Does the AI search the attachments to the entries? For example, if we attach a PDF to a Q&A pair, is it also searching the PDF?
A: At this time, AI does not search through attachments on a Q&A pair unless the document is in the Document Library and is not filtered out.

Q: Is it possible to link the Ask feature in Responsive to an in-house AI tool/LLM?
A: At this time, Ask only exists within Responsive or as a desktop app. If you have LookUp, it also lives within Word and Excel. Learn how to download the desktop app