Hello everyone! My name is Trey and I manage the HubSpot product support community. I came here to get some feedback and inspiration on clever ways we can use AI to improve the product community experience. Some of the ideas I’ve had so far are: 1. Use AI to scrape all of our product ideas, and find duplicates and merge them. 2. To use AI to automatically tag in other members that have expertise in that topic/subject. 3. Use AI to help those submitting questions to build out a quality question (steps to reproduce, browser/OS, feature/hub they are using, etc) to make it easier for our superusers and support team to help them. Thoughts on any of the above? Any other cool ideas you can think of either from your AI experience or what you desire in a community/forum experience? Personalization is definitely on our radar too, but that’s coming later this year. Looking forward to learning from you all! --- This post was migrated from connect.com and was originally published at an earlier date.
Thanks for sharing your question Trey ! Tagging here J.D. Chiles Leo Rodman Clifford Lobo and Jesse Bouman any thougts? --- This post was migrated from connect.com and was originally published at an earlier date.
Hi Victor Becerra , Trey This is an excellent initiative. On this topic, here’s one idea that immediately comes to mind. - - Immediate AI’s Response to the questions submitted -– As an example, here’s a process related question that requires some technical adjustments to the system. How to exclude products from MRR (link) Inbuilt AI tools will generate some arbitrary response since it doesn’t understand the specific technical architecture or framework within HubSpot as a solution. If there’s a way to store all FAQ’s, technical documents and other previous responses on this subject and let AI parse this, the AI will automatically generate the 1st response for an initial discussion, while other community members can chime in. This further can be optimized through AL/ML techniques to understand the context better and in coming months, the custom AI built for the product community can be even more accurate. Note - While the scope to enhance the product community through AI is enormous, identifying some of the current pain points would be a good start. I’ve added the idea above since we’re only brainstorming at this time. HTH. Thanks for tagging for my two cents
--- This post was migrated from connect.com and was originally published at an earlier date.
These are great ideas! Auto meta tagging is probably a good idea. This would make it easier to cross reference related posts automatically for the user (“see also” section) and probably would help the post merging algo too --- This post was migrated from connect.com and was originally published at an earlier date.
Clifford Lobo that’s a super clever idea and would really help us meet our customers where they’re at by surfacing the right content at the right time. This is a common problem that successful communities or companies have which is the fact that there is SO much information and help available. Being able to dial in what we show to our customers could really help streamline their ability to get what they need QUICKLY to be successful with the product. Leo Rodman, this would be super helpful, too. One thing we want to tackle is content archival, but we rely on users to add relevant tags, which they don’t always do. If AI could analyze and add the relevant tags, this would not only help us organize our content but also prepare us for personalization efforts later in the year, as we’ll know that all of the content is tagged and can be surfaced based on our personalization algorithms. We also have an out-of-the-box component that shows “recommended content” based on their search, but admittedly that logic is not the best. Replacing the OOTB logic with AI logic could make this MUCH more valuable, especially for those customers that come through search (anonymous users) and we don’t have any information to personalize their experience with. Their search terms could be the data we use to personalize their results. Love this! Thank you both so much for the feedback and inspiration. 
--- This post was migrated from connect.com and was originally published at an earlier date.
Hi there. I think auto-generating links for resources or articles based on keywords or topics would be helpful to start discussions within the community. --- This post was migrated from connect.com and was originally published at an earlier date.
Thanks Trenika Blevins ! I appreciate all of the feedback. I’ll be sure to circle back here once we have some prototypes in place and would love to hear from your collective expertise. 
--- This post was migrated from connect.com and was originally published at an earlier date.
Trey I think that the most valuable thing that you can do is to anonymize the data as much as possible then, categorize the responses, put them into spreadsheets, and upload them to a model so that you can analyze the written data to find the common themes and threads. Then cross reference that data against the various products, time/period that the products were posted about, and any other profile information that you’re able to gather on a consistent basis across members. This would open up an entirely new window of possibilities for you to be able to very specifically message, target, and speak to those particular types of members based on the products that they purchase and are interested in. Lastly, as just mentioned, you were able to scrape the data to determine product interest and compare that against current product purchased to find out what initial products end up leading to the next purchase, as well as, what specifically leads them to make that purchase based on the summarized written responses. Now you have a detailed and personalized email nurture campaign to use when someone purchases their initial product, that will have a much higher success rate of getting them to purchase additional related products based on the trend of current customers who follow the same pattern. Using the period products were purchased compared against the other data will tell you when you should be running your promotions most often or when they’re most effective. You can verify by comparing to sales revenue on those products at that time. You can go much further by also taking into account the digital activities surrounding the second purchase to also tell you what content from your current offerings to show, as well as where to direct them from your email nurture. --- This post was migrated from connect.com and was originally published at an earlier date.