Community in the Age of AI - CourtneyR.dev's 2nd Brain

Community in the Age of AI 🔗 **Source:** [Community in the Age of AI](https://www.youtube.com/watch?v=njWP4gWiopA) > [!note]+ Document Notes > tags: [developer-community, AI-integration, community-strategy] > author: Developer Community Speaker > source: DevRel Conference > reference: https://www.youtube.com/watch?v=njWP4gWiopA > date created: 2024-12-19 09:30 > > # 🏷️ Title: Community in the Age of AI >  ## 🧰 Action Inbox - [H1] Implement AI bot for simple support questions — reduce noise in community spaces #devrel - [H2] Launch mentorship program to connect experienced/new developers — build peer connections #community - [H3] Balance AI automation with human connection — resist urge to automate everything #strategy ## 🧭 Table of Contents - [📝 Summary](#-summary) - [📖 Main Content](#-main-content) - [🗒️ Notes](#️-notes) - [📚 References](#-references) - [🗣️ Paraphrase](#️-paraphrase) - [📰 Newsletter](#-newsletter) - [✅ Action Items](#-action-items) - [🧪 Hypotheses](#-hypotheses) - [📊 Methodology](#-methodology) - [❓ Questions](#-questions) - [🔮 Future Research](#-future-research) - [💡 Implications](#-implications) - [🧠 Learning Styles Assessment](#-learning-styles-assessment) - [📊 Accessibility Assessment](#-accessibility-assessment) - [🌍 DEIB Assessment](#-deib-assessment) - [✅ Logical Validity](#-logical-validity) - [🛡️ DARVO Analysis](#-darvo-analysis) - [🎯 Conclusion](#-conclusion) ### 🔗 References - [HubSpot Developer Community](https://developers.hubspot.com/) - [Community.hubspot.com](https://community.hubspot.com/) ## 📝 Summary - HubSpot's DevRel team is redesigning their developer community strategy to balance AI automation with human connection - They're implementing "Sprocky," an AI bot to handle repetitive support questions and free up space for deeper peer connections - The approach focuses on amplifying quieter community voices while using AI to filter out noise from "squeaky wheels" - Key metrics include peer-to-peer connections, event attendance, and AI integration effectiveness - Timeline spans 18-24 months with phases including Slack 2.0, mentorship programs, and agent partnerships ## 📖 Main Content The speaker, a DevRel professional at HubSpot, presents their strategy for navigating the intersection of AI and developer community building. HubSpot operates a CRM platform with developer tools, integrations, and a CMS, serving both traditional developers and emerging "vibe coders" who use AI to generate functional code. Their current challenge involves a 10-year-old developer Slack with 25,000+ members but low engagement and high churn. The space is flooded with unanswered questions, creating poor experiences for everyone. Rather than purely scaling human moderation, they're implementing AI strategically to handle routine queries while preserving space for meaningful human interaction. The core philosophy centers on having "AI amplify but not replace human potential." This means using automation to reduce noise and support administrative tasks while prioritizing peer-to-peer connections, mentorship, and community leadership development from historically quieter voices. ## 🗒️ Notes #### ⏱️ [00:00](https://www.youtube.com/watch?v=njWP4gWiopA&t=0) — [H1] Introduction and Industry Excitement - Key idea: Speaker feels most excited about tech since the API-as-product era - Evidence: "I haven't felt this excited about tech since that era and that's saying a lot" #### ⏱️ [02:30](https://www.youtube.com/watch?v=njWP4gWiopA&t=150) — [H2] HubSpot Context and DevRel Role - Key idea: HubSpot is a CRM platform with developer integrations and CMS offerings - Evidence: "We facilitate marketing campaigns, sales campaigns, operations and services" #### ⏱️ [04:15](https://www.youtube.com/watch?v=njWP4gWiopA&t=255) — [H1] Core Philosophy: Amplify Not Replace - Key idea: AI should enhance human potential rather than substitute for human connection - Evidence: "AI needs to amplify but not replace human potential" #### ⏱️ [06:45](https://www.youtube.com/watch?v=njWP4gWiopA&t=405) — [H2] Evolving Definition of Developer - Key idea: Developer roles are expanding beyond traditional coding to include business users with AI tools - Evidence: "We're starting to see more and more business users starting to leverage AI... for generating code" #### ⏱️ [08:30](https://www.youtube.com/watch?v=njWP4gWiopA&t=510) — [H3] Wonder as Developer Defining Factor - Key idea: Developer identity should include anyone inspired by moments of wonder and discovery - Evidence: "If you're really into those moments of wonder and inspiring them in yourself and others, then I consider you to be a developer" #### ⏱️ [10:20](https://www.youtube.com/watch?v=njWP4gWiopA&t=620) — [H1] Current Community Challenges - Key idea: Large Slack community suffers from high churn and walls of unanswered questions - Evidence: "25,000 plus members... walls and walls of unanswered questions and it's a bad experience for everybody" #### ⏱️ [12:50](https://www.youtube.com/watch?v=njWP4gWiopA&t=770) — [H2] Strategy: Less Noise, More Connection - Key idea: Focus on reducing broadcast communication in favor of peer-to-peer developer connections - Evidence: "Less noise, less broadcasting and more connection between developers building on HubSpot" #### ⏱️ [14:40](https://www.youtube.com/watch?v=njWP4gWiopA&t=880) — [H2] Mentorship Program Launch - Key idea: Ecosystem mentorship program paired 100 people for career development - Evidence: "We launched our ecosystem mentorship program this year... paired a hundred folks" #### ⏱️ [16:30](https://www.youtube.com/watch?v=njWP4gWiopA&t=990) — [H3] Redefining Community Leadership - Key idea: AI can help quieter community members shine by reducing dominance of "squeaky wheels" - Evidence: "Use AI to allow those sort of squeaky wheels to find and maybe interact with a bot... and make more space for the folks who might be historically a little quieter" #### ⏱️ [18:15](https://www.youtube.com/watch?v=njWP4gWiopA&t=1095) — [H2] Key Success Metrics - Key idea: Peer-to-peer connections are the most important community metric to track - Evidence: "The tried-and-true metric... has been peer-to-peer connection" #### ⏱️ [20:45](https://www.youtube.com/watch?v=njWP4gWiopA&t=1245) — [H2] Introducing Sprocky AI Assistant - Key idea: AI bot named Sprocky will handle routine questions and measure unblocking effectiveness - Evidence: "Sprocky is about to get a makeover and get... AI steroids and become incredibly helpful" #### ⏱️ [23:20](https://www.youtube.com/watch?v=njWP4gWiopA&t=1400) — [H3] Multi-Phase Implementation Timeline - Key idea: 18-24 month rollout starting with Slack 2.0 in September, expanding to mentorship and partnerships - Evidence: "We're hoping that our Slack 2.0 will go live by early September" #### ⏱️ [26:10](https://www.youtube.com/watch?v=njWP4gWiopA&t=1570) — [H2] Agent Partnerships Strategy - Key idea: HubSpot wants to work with AI agents as part of their developer audience - Evidence: "We also want to start working with actual agents. So I think agents are going to become part of our audience" #### ⏱️ [28:30](https://www.youtube.com/watch?v=njWP4gWiopA&t=1710) — [H1] Ultimate Vision: Minimal Friction Innovation - Key idea: When developers can build with minimal friction, it creates positive business flywheel effects - Evidence: "When developers are able to build cool stuff with minimal friction, the flywheel just spins and spins much more quickly" ## 📚 References - [HubSpot Developer Platform](https://developers.hubspot.com/) - [Mentor Loop Tool](https://mentorloop.com/) - [Common Room Community Management](https://www.commonroom.io/) - [Community Commitment Curve Research](https://www.communityroundtable.com/) ## 🗣️ Paraphrase - HubSpot is redesigning their developer community to use AI for handling routine questions while preserving human connections for deeper engagement and mentorship - The strategy involves implementing an AI assistant called "Sprocky" to reduce noise in their 25,000-member Slack community and create space for quieter voices to emerge as leaders - Success will be measured through peer-to-peer connections, event participation, and AI effectiveness rather than traditional volume metrics, with a multi-phase rollout over 18-24 months ## 📰 Newsletter **Community Building in the AI Era: Lessons from HubSpot's DevRel Strategy** HubSpot's developer relations team is taking a thoughtful approach to integrating AI into their community strategy. Rather than automating everything, they're using AI strategically to handle repetitive support questions while preserving space for meaningful human connections. Their "Sprocky" AI assistant will filter out routine queries, allowing community members to focus on deeper discussions and peer mentorship. This approach particularly aims to amplify historically quieter community voices while reducing dominance from more vocal members. Key takeaways for community builders: measure peer-to-peer connections over volume metrics, use AI to reduce noise rather than replace human interaction, and create structured mentorship programs to build lasting relationships. The multi-phase rollout includes Slack improvements, expanded mentorship programs, and even partnerships with AI agents as community participants - recognizing that the definition of "developer" continues to evolve in the AI era. Resources: [HubSpot Developers](https://developers.hubspot.com/), [Community Management Best Practices](https://www.communityroundtable.com/), [DevRel Resources](https://developerrelations.com/) ## ✅ Action Items #### DevRel - Developer advocacy: Launch AI-powered mentorship programs with algorithmic pairing and success tracking - Education: Create workshops on AI-assisted development workflows for traditional and emerging developer roles - Developer experience: Implement intelligent bot (Sprocky) to handle routine documentation and API questions - Developer marketing: Position community as connection-focused rather than support-focused in messaging - Community management: Redesign Slack architecture to surface quieter voices and reduce noise from frequent posters #### OSPO - Policy & Governance: Establish guidelines for AI bot interactions and community moderation protocols - Developer Support: Create escalation paths from AI bot to human experts for complex technical issues - Community Engagement: Partner with AI agent developers as emerging community audience segment #### Community - Maintainers: Develop peer-to-peer connection metrics and tracking systems for community health - Contributors: Design onboarding flows that balance AI assistance with human mentor matching - Ecosystem: Build partnerships with AI development tools and platforms for integrated experiences #### WordPress - Community: Adapt HubSpot's mentorship program model for WordPress contributor onboarding and retention - Support: Implement intelligent triage bots for support forums while preserving human escalation paths - Documentation: Use community question themes from AI interactions to inform documentation priorities ## 🧪 Hypotheses - AI bots handling routine questions will increase meaningful peer-to-peer interactions in developer communities - Mentorship programs with algorithmic matching will produce stronger career outcomes than self-selected pairings - Communities that balance AI automation with human connection will retain members better than purely automated or purely human-moderated spaces - Quieter community members will emerge as leaders when AI reduces noise from more vocal participants ### 📊 Methodology - A/B test community spaces with and without AI bot integration, measuring peer interaction rates - Longitudinal study tracking career progression of mentorship program participants versus control groups - Sentiment analysis of community conversations before and after AI bot deployment ### ❓ Questions - How do we maintain community authenticity when AI handles increasing portions of member interactions? - What happens to community knowledge sharing when routine questions get answered privately by bots? - How do we prevent AI assistance from reducing the learning opportunities that come from struggling with problems? ### 🔮 Future Research - Long-term impact of AI-mediated community interactions on developer skill development - Optimal balance points between AI automation and human connection in technical communities - Evolution of community leadership roles as AI handles more administrative and support functions ### 💡 Implications - Practice: Community managers need new skills in AI tool management and human-AI workflow design - Policy: Communities require clear guidelines about AI bot capabilities, limitations, and escalation procedures - Security: AI bots handling community questions create new attack vectors and data privacy considerations - Education: Community members need training on effective AI assistant interaction alongside peer networking - Funding: Investment in AI tooling must be balanced with continued investment in human community programs ### 🧠 Learning Styles Assessment Reading level estimate: 12th grade #### 👀 Visual learners - Present: Slide deck with community architecture diagrams and metrics - Improve: Add community interaction flow charts and AI bot decision trees #### 🔊 Aural learners - Present: Conference talk format with Q&A discussion - Improve: Podcast-style deep dive on implementation details #### 🗣️ Verbal learners - Present: Clear definitions of developer roles and community terminology - Improve: More analogies comparing AI integration to familiar community patterns #### 🧩 Kinesthetic learners - Present: Mentions of hands-on mentorship program and practical implementation - Improve: Workshop format with actual bot configuration exercises #### 🧠 Logical learners - Present: Clear timeline, metrics framework, and phased implementation plan - Improve: Cost-benefit analysis and ROI calculations for AI investments #### 🧑‍🤝‍🧑 Social learners - Present: Focus on peer-to-peer connections and mentorship relationships - Improve: Group discussion guides for implementing similar programs #### 🧘‍♀️ Solitary learners - Present: Individual reflection on community leadership and career growth - Improve: Self-assessment tools for community contribution styles #### 🌿 Naturalistic learners - Present: Organic community growth metaphors and ecosystem thinking - Improve: More connections to natural community formation patterns ### 📊 Accessibility Assessment #### Screen reader semantics - Present: Clear heading structure and descriptive content - Improve: Add alt text for slide content and visual community diagrams #### Keyboard navigation - Present: Standard video player controls - Improve: Transcript navigation and searchable content structure #### Contrast and readability - Present: Standard video format meets basic requirements - Improve: High-contrast transcript and accompanying materials #### Text resizing and reflow - Present: Video content scales appropriately - Improve: Responsive transcript and note-taking templates #### Media and captions - Present: Video with audio content - Improve: Professional captions and downloadable transcript #### Forms and interactions - Present: Simple Q&A format - Improve: Accessible feedback forms for community strategy input ### 🌍 DEIB Assessment #### Cultural breadth - Present: Tech industry perspective primarily from US-based company - Improve: Global community management examples and cultural considerations #### Inclusive language - Present: Uses inclusive terms like "folks" and avoids gendered assumptions - Improve: Address accessibility barriers in community participation #### Neurodiversity support - Present: Acknowledges different communication styles (quiet vs vocal members) - Improve: Specific accommodations for neurodivergent community participants #### Representation - Present: Single speaker perspective - Improve: Multiple voices from different community roles and backgrounds #### Psychological safety - Present: Emphasizes supporting quieter community members - Improve: Explicit discussion of harassment prevention and inclusive moderation ### ✅ Logical Validity #### Strong reasoning - Deductive strength: Clear logical progression from problem identification to solution design - Inductive strength: Good use of specific examples (25k Slack members, mentorship program results) to support general claims - Causal reasoning: Establishes clear cause-effect relationships between AI automation and human connection outcomes - Analogical strength: Effective --- ## Full Transcript [00:08] All right. So again, thank you so much [00:10] for that intro. It was really nice. I [00:12] wrote it. So [00:14] um and if anyone is feeling also [00:17] conflicted about the tech industry [00:18] overall or AI, I am happy to talk about [00:21] those things. But I am simultaneously so [00:23] excited about AI and about the tech [00:26] industry these days. Um I've been around [00:29] since the eras of you know when APIs as [00:32] products like you know toilet and send [00:33] grid and a lot of other companies um [00:35] started to get funding and started to [00:37] become crucial infrastructure for all [00:40] kinds of you know amazing apps that we [00:42] know and love today. And so I haven't [00:44] felt this excited about tech since that [00:46] era and that's saying a lot. So um I [00:49] want to talk to you a little bit about [00:51] how I'm sort of interpreting the [00:53] potential for AI when it comes to [00:55] building out developer community. Um I [00:57] do work at HubSpot and uh my colleague [01:00] Nikita will be moderating the panel at [01:02] the end of the day too. So it's nice to [01:04] have some HubSpot representation here. [01:05] For those of you who don't know, HubSpot [01:07] is a CRM and business platform that [01:09] businesses can use to manage all aspects [01:11] of their business. Um really driven by [01:14] the core uh data around their contacts. [01:17] So you know if you want to run your [01:19] marketing campaigns, run your sales [01:20] campaigns, uh run your operations and [01:23] services, we facilitate all of that. Um [01:26] Devril sits on the side of allowing [01:28] folks to build integrations, [01:30] customizations. Um we also have a CMS [01:33] offering. So we have a lot of front-end [01:35] developer community which I'll talk a [01:37] little in a moment. Um but we're doing a [01:39] lot of interesting things on Devril and [01:41] um I think we're very excited and also [01:44] cautiously you know careful about all [01:47] all of these changes that are happening [01:48] right now. Um so let's get into it. What [01:52] I want to talk about today [01:57] back slide. [01:59] So essentially what I'm trying to do is [02:01] you know talk to you today about how [02:03] we've been trying to navigate that [02:05] intersection of AI and dev community. I [02:07] want to tell you a story about how we're [02:08] resisting the urge to automate [02:10] everything. Even though that's been a [02:11] big ethos in HubSpot for a long time. [02:14] The more you can automate, the more time [02:15] you have to do really meaningful work. [02:17] But we're trying to balance that [02:19] automation that AI can really accelerate [02:21] with um the ability to allow folks to [02:24] connect even more deeply. Um so our [02:27] philosophy on Devril, I apologize this [02:29] slide got a little wacky. It's my fault [02:31] for sending last minute slide edits um [02:34] like an hour before this. So I get to [02:36] see everything, but our our philosophy [02:38] on dev community is really that AI needs [02:40] to amplify but not replace human [02:42] potential. Um so I'm going to talk about [02:44] what we're working on, what's still a [02:46] little bit messy, but mainly our vision. [02:48] Um I've put together this year and a [02:51] half to two year vision. We're not [02:52] looking like fiveyear visions these days [02:54] because AI is changing so rapidly. Um, [02:57] but I'm making a lot of assumptions that [02:58] I hope are going to be true that will [03:00] help us balance this, you know, AI [03:02] functionality, also AI enablement [03:05] enablement as well as that human [03:07] connection. That's going to be in my [03:08] opinion more important than ever. Um, so [03:11] it's not really a HubSpot story [03:12] exclusively. What I'm really hoping [03:14] today is that I'm going to share a [03:15] framework of thinking that folks here [03:18] can learn from, adapt, and use as [03:20] they're as needed for their own [03:22] communities, developer communities. And [03:25] um let's get started. So I'm going to [03:28] take a step back. I think most of us are [03:30] pretty comfortable and familiar with I [03:32] guess what I would call for lack of a [03:34] better term traditional developers that [03:36] are working in the code day in day out. [03:38] Um, but we're starting to see, [03:39] especially, I think all of us are [03:41] starting to see this across our [03:42] companies and platforms, but in HubSpot, [03:44] we're specifically starting to see more [03:46] and more business users starting to [03:48] leverage AI, not even just for enhanced [03:50] workflows and, you know, quicker um, [03:53] automations, but also for generating [03:56] code that does stuff on HubSpot. And so [03:59] we're in this really interesting space [04:01] right now where we have uh evolving [04:04] traditional developers, you know, these [04:06] folks that we're calling AI engineers [04:07] that are somewhere in the middle of [04:09] these vibe coders on the business side [04:12] and our traditional developers that are [04:14] looking more and more at things like [04:15] infrastructure, scalability, um cleaning [04:18] up messy code generated by tools like [04:21] Lovable. Um and you know, shout out to [04:23] Lovable. I did build my website on them, [04:25] so I'm big fan. But um and there's going [04:27] to be all kinds of emerging roles coming [04:29] up in the next few years that we can't [04:31] possibly anticipate. Um I'm hoping [04:33] there's going to be more and more AI [04:35] ethsists, hence the internal conflict. [04:37] But um we're going to start to see our [04:40] our world shift quite a bit. But I am [04:43] confident and hopeful that developer the [04:46] term developer will still be a word, but [04:48] we'll just have a broadened definition [04:51] and a different set of sort of table [04:53] stakes requirements. Um, [04:57] with that I want to talk about our [04:58] strategy on dev community. It's really [05:01] about less noise, less broadcasting [05:04] and more connection between developers [05:07] building on HubSpot. And just as a [05:09] caveat, I will be using the word [05:11] developer to sort of encapsulate all of [05:13] those audiences I shared in the last [05:15] slide. And I did want to also add [05:17] something. I don't know if for folks who [05:18] were at Ricky's talk this morning. Um my [05:21] sort of like community-minded heart [05:24] loves the concept of wonder as being a [05:26] defining factor of a developer. So if [05:29] you're really into um those moments of [05:31] wonder and inspiring them in yourself [05:33] and others, then I I consider you to be [05:35] a developer. So that's just my aside. [05:37] Um, so what we're doing is rebuilding [05:40] our community space that we are running [05:42] on dev community team right now and [05:44] really focusing it around connection. [05:46] And that's something I've been wanting [05:47] to do in various ways for years. Um, and [05:51] we've been kind of subject to [05:54] HubSpot is a large company. We have I [05:57] think almost 10,000 employees at this [05:59] point. Um, and so there's a lot of [06:01] different inputs coming from different [06:03] places into our community spaces. The [06:06] dev slack is something my team runs. Uh [06:09] it has existed for the last 10 years. Um [06:12] and it on paper if you join it you will [06:15] see channels that have 25,000 plus [06:17] members. Very deceiving kind of metric [06:20] because the folks who are actually [06:21] active in there are much much lower just [06:24] a percentage of that. And what we see is [06:26] a ton of churn and turnover. folks get [06:29] referred to that space by our support [06:31] team who are amazing but can't always [06:34] handle the nuance complexities that our [06:36] developers come in asking. So they'll [06:38] send them to our community spaces and [06:41] we'll what's happened is because we're a [06:43] really small team, we have not had the [06:46] ability to create programs and [06:48] structures to um really um shape the the [06:52] way that that Slack interaction is [06:53] going. So you see walls and walls of [06:55] unanswered questions and it's a bad [06:57] experience for everybody. Um, we've been [06:59] trying all these different approaches to [07:00] fix that. But what we're doing now is [07:02] really looking around how AI can help [07:05] um, speed up and facilitate some of the [07:08] some of the repeatable tasks, the [07:11] ongoing support questions to make space [07:13] for real contact uh, connection. And [07:16] we're looking at also helping our [07:19] community grow in this AI minded future. [07:22] So, we're creating community programs. [07:24] We launched our ecosystem mentorship [07:26] program this year. [07:27] um where we're paired a hundred folks [07:30] with, you know, folks who have been seen [07:31] a lot of success in the HubSpot [07:33] ecosystem with folks who are trying to [07:34] upskill and uplevel in their own [07:37] journeys on HubSpot. Um we want to lean [07:39] into that even more and have more time [07:41] to do that. And fundamentally what we're [07:43] trying to do is to [07:46] change the definition of leadership in [07:48] our community. I think historically [07:50] because of these sort of barriers to [07:52] entry and the amount of noise, it's been [07:54] kind of the squeaky wheels who have been [07:56] the kind of most influential and loudest [07:59] and the folks been able to get what they [08:00] need from us. But we want to use AI to [08:04] allow um those sort of squeaky wheels to [08:08] find and maybe interact with a bot a [08:10] little bit more and less vocally in our [08:13] community and make more space for the [08:15] folks who might be historically a little [08:17] quieter. um and really give them the [08:19] chance to shine. [08:21] Before I get into the tactics [08:23] specifically, I want to talk about how [08:24] we're going to be, you know, measuring [08:26] everything. And I would say the first uh [08:29] set of metrics around community growth. [08:31] I don't want to be thrown off by the [08:33] word growth. What I mean by growth here [08:34] is really growth of the metrics that I [08:37] think are most important in community. [08:38] And the triedand-true metric for uh the [08:41] work I've been doing for the last 12 to [08:42] 15 years has been peer-to-peer [08:44] connection. So, as much as we can [08:46] measure um non-hubspotters in our case [08:49] because that's the way our community [08:51] works. In some cases, your company is [08:52] going to have your advocates in the [08:53] spaces all day answering the questions [08:55] and that's one model. Um and I think [08:57] that works to a certain size of company, [08:59] but at a um the point that we're at for [09:02] a twoerson dev community team um we [09:05] don't have the time to answer those [09:06] questions all day. And so, we want to [09:08] make it so that our developer community [09:11] feels inspired to help each other out. [09:13] So peer-to-p peer connection is a huge [09:14] metric for us. Um and we want to see [09:18] more impact across groups. And when I [09:20] talk about event attendance in our [09:22] community, I'll also include like [09:24] virtual events, um inerson events. Um I [09:27] don't know if anyone was at Erica [09:28] Hansen's talk earlier today where she [09:30] talked about building community with [09:32] Google and and Shutterfly or Flutterfly. [09:34] So not Shutterfly, it's a different [09:35] company. Um, but she referenced the [09:38] community commitment curve. And I wish I [09:39] had that slide up here right now, but [09:41] essentially the premise of the community [09:43] commitment curve is all of these [09:44] different milestones that lead folks to [09:47] in a community. If they're giving more [09:50] time to a community, then they're more [09:52] committed. And that generally results in [09:54] better business outcomes. And so to me, [09:56] events like this are part of, you know, [09:59] in our case is maybe an overall moment [10:01] of catalyzing our ourselves around our [10:04] community ident or our career identities [10:06] and meeting our peers around this [10:08] career. And so events for us would be [10:10] things like, you know, let's get folks [10:11] together around the mentorship program [10:13] in person and talk about um, you know, [10:16] what it's meant to them, deepen those [10:18] relationships, and it will just [10:19] influence that kind of journey even more [10:21] moving forward. Um, I think I may have [10:24] skipped a slide. Sorry, not used to [10:27] these clickers. Um, so let me talk a [10:30] little bit about AI integration. This is [10:31] a whole new set of metrics for us. Um, [10:33] what we're going to be hoping for is [10:35] this new tool that I'll talk about in a [10:37] moment. His name is going to be Sprocky. [10:40] Um, our logo is a sprocket. And [10:43] Sprocky's actually been around in [10:45] HubSpot's history as a kind of clippy [10:48] adjacent assistant in various spaces. [10:51] Um, but uh, Sprocky is about to get a [10:53] makeover and get, I don't know, AI [10:56] steroids and become incredibly helpful. [10:58] Um, and so we're going to be looking at [11:00] how how much Sprocky's being used, how [11:03] many folks are getting unblocked more [11:05] quickly, um, how many, you know, like [11:08] the accuracy of of Sprocky. And [11:10] fundamentally, these two areas of of uh, [11:13] measurement are going to result in [11:14] ecosystem impact. So more retention both [11:17] across developers and customers, more [11:19] marketplace growth and quality, and [11:21] higher developer trust, which is I think [11:23] probably one of the most important [11:24] metrics for development. [11:27] And so I'm going to quickly go through [11:30] the phases in the timeline that we're [11:31] looking at for this. Right now, we're in [11:33] phase zero, and I'm here talking to you [11:34] about it already, which is very [11:36] ambitious. Um, but really focusing on [11:38] building out um that Slack experience, [11:40] improving it, getting that bot in there. [11:43] Um, and the bot itself is going to be, [11:46] let's see, I think this might actually [11:47] be the next slide. I'll go into that [11:48] after. But, um, really getting that, you [11:51] know, tightened journey, um, solid for [11:55] our community space, our developer Slack [11:57] in this case. Um, from then from there [11:59] all these phases kind of bleed into each [12:01] other, but we're going to lean even more [12:03] into our mentorship scale it um pull out [12:06] some uh specific AI um themed uh I guess [12:11] tracks for the mentorship program, do [12:13] more events and facilitate more peer [12:16] learning. From there, we're going to um [12:18] actually this is a call out to this [12:20] community for folks who are building [12:22] anything that would like integrate with [12:24] HubS benefit from HubSpot's data. I've [12:27] already had some conversations at lunch [12:28] with some folks who are building out [12:30] some agentic experiences that would [12:31] benefit from CRM data. Um we want to [12:34] build more partnerships with those [12:36] folks. So yes, historically we've worked [12:38] with individual developers and some um [12:40] agencies, but uh we also want to start [12:43] working with actual agents. So I think [12:46] agents are going to become part of our [12:48] audience as we grow. Um and finally, you [12:51] know, all of that is going to lead to [12:52] what I'm hopeful will be much more [12:54] intelligent expansion. So this bot that [12:56] we're creating, we'll be able to roll it [12:58] out into different spaces. It is a part [13:00] of an overall um kind of approach that [13:03] we're taking to our docs. Um a lot of [13:06] stuff is changing um in the next few [13:08] months for the HubSpot developer [13:10] experience that I'm very excited about. [13:11] So all the stuff ties together and what [13:13] I'm hoping is that we can use not only [13:15] AI and the bot to improve the unblocking [13:19] of de of devs as they're building and [13:21] leave more space for sort of like deeper [13:24] relationship based connections but also [13:26] use the the questions that are incoming [13:28] to inform our product and to um you know [13:32] in inform the roadmap and allow us to [13:34] have more real-time insights on those [13:36] things. [13:37] timeline. It's probably a little [13:39] redundant, but we are um hoping that our [13:43] Slack 2.0 will go live by early [13:45] September. Um from there, we're going to [13:47] roll out an AI powered mentorship pilot [13:50] and uh start leaning more and more into [13:52] developer gatherings, async learning [13:54] spaces, more virtual and inerson events, [13:57] and then really lean into those [13:59] partnerships I was discussing in uh [14:01] 2026, and so on. Um, I wanted to show [14:04] that there is a little bit of momentum [14:06] we need to hit here at this moment [14:07] because I do think that between now and [14:10] a year and a half or two years from now, [14:12] um, we're going to start to see the [14:13] establishment of foundations and [14:15] expectations of how of this messy middle [14:17] between these tools that are, you know, [14:20] vibe coding tools and actual like [14:22] scalable production tools um, and [14:25] outcomes are going to meet in the middle [14:27] and we'll we'll understand what that [14:28] looks like more. [14:30] Um, and so I'm going to deepen do a bit [14:33] of a deeper dive into uh phase one. So [14:36] we're really uh we're I'm really really [14:38] proud of mentorship program. It's [14:40] probably one of the things I'm most [14:41] proud of that I've built in as part of [14:43] the HubSpot developer community work [14:44] that I've done. Um we've been able to [14:47] really build some tight connections with [14:49] people that are very intentional and [14:51] really focused on allowing folks to [14:53] build careers. And so we want to take [14:56] what we've learned from it so far, lean [14:57] into a tool that we're using called [14:59] mentor loop which does have some AI [15:01] functionality itself um some algorithmic [15:04] machine learning tooling to pair folks [15:06] together as well as more and more AI [15:08] features in the tool itself. So it's you [15:10] know when I say AI it's this very big [15:12] thing and I think in some cases it's [15:13] applied to the tooling itself and it's [15:15] also applied to us facilitating learning [15:18] around AI. We're going to lean more into [15:20] gatherings and we're going to lean more [15:22] and more into those feedback loops that [15:24] I think are kind of missing today um on [15:26] the scale that we need them. And um from [15:29] there, here's the team that's pulling it [15:31] together. Um we have someone focused on [15:34] strategy, that would be me. I have an [15:35] amazing colleague. So, we're a two [15:37] person team, like I said before. Her [15:38] name is Jennifer Nixon, and she is [15:40] leading our execution and operations of [15:42] all of these great things. And we have [15:45] our extended crew. We have the developer [15:46] advocates that can help out with a lot [15:48] of this, benefit from a lot of this. Uh [15:50] contractors that are helping us build [15:52] out Sprocky's functionality and some of [15:54] these partners that we want to build the [15:56] deeper relationships with. And finally, [15:57] I'm sort of seeing Sprocky like this is [15:59] where I hesitate a little bit because by [16:01] no means do I ever want AI to replace a [16:02] human, but Sprocki is kind of going to [16:05] be on our team. Um we're going to be [16:07] leaning into this tool to really help us [16:09] accelerate all of these other things. [16:12] And then this is my sort of uh closing [16:14] slide. What we really are aiming for [16:16] with all of this is the ability to sort [16:19] of facilitate more innovation. You know, [16:22] when developers are able to build cool [16:24] stuff with minimal friction, the [16:27] flywheel just spins and spins much more [16:28] quickly. So, my hope is that if we get [16:30] the balance right of where AI is [16:32] automating, AI is unblocking folks with [16:34] answers to quick support questions, then [16:37] it's just going to open up the time to [16:39] to work on the stuff that matters. And [16:42] um our community will improve. will have [16:44] more ability for, you know, maybe [16:46] typically quieter voices to to be heard [16:48] because they're going to be funneled [16:50] into the relevant conversations at the [16:52] right moments based on their kind of bot [16:54] interactions. And finally, the the [16:57] bigger vision is to have all of this [16:59] drive career growth. Um career growth, [17:01] of course, in our case based on, you [17:03] know, the skills that it requires to [17:04] help customers uh be successful in [17:06] HubSpot. But ultimately um that's like [17:10] you know something that you can apply to [17:12] your ultimate business goals as well. [17:15] Um I don't know if we have time for [17:16] questions um but if [17:19] >> we do have time for a few questions [17:21] >> raise your hand if you have any [17:22] questions. [17:25] >> I am very curious about kind of like if [17:27] anyone's been playing with tools with [17:29] their community that have like you know [17:30] resulted in embarrassing hallucinations [17:32] or incorrect answers. I think that's [17:35] some of the messy stuff that I didn't [17:36] talk about. Um so if anyone wants to [17:39] chat with me about before or after [17:41] that's great. [17:46] >> Hi um that was a great talk. Uh I had a [17:50] question about your community strategy [17:51] around um you know when customers are [17:53] answering questions to other customers. [17:56] How do you how do you deal with like [17:57] moderation and you know someone giving [18:00] like an answer that's not what you would [18:03] want them to give to another customer [18:04] and things like that? Like what's your [18:06] strategy? Yeah, that's a really good [18:07] question and I think it's one of the [18:10] balances you need to strike and sort of [18:12] make a little bit of peace with when you [18:14] are driving customers to community [18:17] spaces. And so that includes some [18:19] caveats before they post. That includes [18:22] some, you know, warnings that you're not [18:24] getting the official answer, you're [18:25] getting an expert answer. Um, we do have [18:28] other spaces that facilitate facilitate [18:31] community support at HubSpot. I [18:33] specifically talked about the dev Slack [18:35] today, but we do have [18:36] community.hubspot.com, [18:38] which is where the majority of support [18:40] questions are funneled and in those [18:41] cases, we have pretty amped up champions [18:44] programs that are there, but those [18:45] spaces are largely for the the customer [18:48] focused questions. um our sort of [18:51] mandate on Devril is to help support [18:53] developers and we've created these [18:55] alternate spaces um because there wasn't [18:58] a lot of frontline support for or even [19:00] community support at the scale we [19:02] needed. Um so we have a dedicated space [19:05] for them there. Um but yeah, that's a [19:07] really good point. Like one of the [19:09] things we just did was extract an entire [19:12] um uh basically uh what's the word I'm [19:15] looking for? Export of our uh developer [19:17] Slack over the last 10 years. did some [19:20] anal analysis of it and were able to [19:23] identify like this no longer applies. [19:25] This is true. We can use this to um you [19:28] know support some content that is being [19:30] developed now. Um so it's an internal [19:32] and an external thing. Um I think the [19:34] caveats are really important when folks [19:37] are being sent there. And we also need [19:39] the ability and I've been pushing for it [19:41] internally to be able to send folks to [19:43] actual product experts when the [19:46] community isn't coming through. And [19:47] sometimes the community is not going to [19:48] know because they don't have access to [19:49] every aspect of the internal road map. [19:52] Um they don't know everything. And [19:54] sometimes questions start out I'm [19:55] yapping now. But sometimes questions [19:58] start out um with one qu one sort of [20:01] like the question is coming from one [20:02] place but it actually is about a bug or [20:05] the other way around. A bug is something [20:06] else altogether. So, it's a combination [20:09] of like giving the caveats, being able [20:12] to triage the questions correctly, and [20:14] using it to inform our um actual default [20:17] answers that hopefully live in our docs [20:20] and hopefully live in our knowledge base [20:21] so that we're reducing the [20:24] inconsistency. So, not a perfect answer, [20:26] very complex issue. [20:29] Hi, I'm uh and I think my question with [20:34] integrating AI into communities is that [20:37] it feels at some times that as AI [20:40] continues to develop, less people go to [20:42] these community spaces to make questions [20:44] that other people can benefit from. [20:46] >> Yeah. [20:46] >> And on the other side, as a developer [20:48] relation like professional, you don't [20:49] get those messages in in your radar as [20:53] they're being asked to like AI. So I [20:55] wonder how are you navigating like these [20:57] two pieces? [20:58] >> Sorry, can you repeat the second part of [21:00] the question? [21:00] >> Um the second part is like if folks make [21:03] questions to like AI assistants. [21:05] >> Yeah. [21:05] >> It's really hard for them to appear [21:07] under your radar or you just don't see [21:10] them appear in your chats. [21:11] >> Yeah. [21:12] >> Um which make it hard to know what are [21:14] some of the barriers that people are [21:15] finding. [21:16] >> Yeah. [21:16] >> So how do you navigate the balance [21:18] between these these components? [21:19] >> Um why am I blanking on the first [21:21] question? I had an answer for it two [21:22] seconds ago. Can you repeat it again? [21:24] I'm so sorry. The first question, [21:26] >> the first question is as we integrate [21:29] more AI, we see less questions and [21:32] people not going to communities to to [21:34] like find [21:35] >> answers to to problems they might find. [21:37] >> Yeah. So, I think we might be a little [21:39] unique in the sense that um we aren't [21:43] necessarily [21:45] creating our community spaces for um a a [21:50] lot a high volume of questions. What we [21:53] want to do is leverage AI to get folks [21:55] unblocked on the simple stuff that [21:57] they're coming in asking over and over [21:59] again. And hopefully that will set the [22:02] stage for more nuance conversations that [22:05] other experts will have opinions on and [22:07] not be just frustrated like oh someone's [22:09] asking the same question again. Um it is [22:12] it is I mean an interesting balance like [22:13] I I'm really curious to see how it kind [22:16] of comes together. Um, but the other [22:18] thing is that when we're building out uh [22:20] the Sprocky bot, we are going to be [22:22] building out ways to pull out frequent [22:25] reports of themes, things coming in. [22:27] That's going to be a crucial part of [22:28] building that out. So, we may lose [22:30] visibility over the specific question. [22:33] Um, although I do anticipate a world [22:35] where we can map those spec specific [22:37] questions back to a user's profile [22:39] somewhere. We do use a tool called [22:41] Common Room for our community [22:43] management. I'm sure lots of you know [22:44] it. And so maybe there's some creative [22:46] stuff we can do there. Um, but uh I [22:49] think it's going to be about those [22:50] themes and being able to [22:53] more directly or indirectly rather [22:56] address the needs of the community based [22:58] on creating content, experiences, events [23:00] around those themes that keep coming up. [23:02] Um, so yeah, I mean we don't there's so [23:04] much we don't know. Um, we're also very [23:06] willing to take a hit on the volume [23:08] metrics um and really u uh look at those [23:11] quality metrics more than anything. [23:16] We good? [23:18] >> All right. Thank you everyone.