AI & community - a tremendous opportunity?

And an invitation to experiment together

Photo by Skyler Ewing

The other day, Marc Fonteijn from the Service Design Show showed me some of the AI experiments he’s developed for his service design community. I have been rather skeptical about the applications of AI in community. AI chat-bots to activate a forum feel like a taster of a dehumanized dystopia. But when speaking to Marc, I recognized that there is tremendous opportunity for communities to play with this new technology and apply it in meaningful ways.

What if AI helps us make the case for communities and networks?

AI offers a solution to one of the biggest challenges facing communities and networks: demonstrating their true value without turning interactions into transactions. Many of us intuitively understand that communities are valuable, but this value often remains hidden beneath the surface. It exists in the quality and quantity of relationships (from casual connections to deep friendships), peer support, personal and professional growth, and collective knowledge (embedded in calls, sessions, notes, and posts). It's found in successful need-offer matches and emerging collaborations.

While incredibly valuable, these benefits remain largely invisible to outsiders. Crucially, much of this value emerges through chance encounters. You might learn something transformative simply because you attended a particular gathering, joined a specific session, or heard a key comment. Or perhaps someone made a life-changing introduction because they happened to see your request. AI could help make this hidden value more visible and accessible.

The value question is crucial, because it directly links to funding and thereby the group's long-term viability. One avenue to fund communities is through philanthropic investment. While many funders claim to invest in systemic solutions (which communities and networks perfectly embody), most ultimately prefer funding tangible and predictable solutions that achieve clear-cut quarterly KPIs. Another avenue is through membership fees. But the moment the community asks for contributions, members also start considering what tangible, short-term outcomes they get from the group. Both funders and paying members want a clarity that most groups can’t provide. And as a result numerous networks struggle to secure sustainable funding, leading to burnout among their leaders.

What if AI provides a pathway for groups to make their value more visible, accessible and predictable? What if this value becomes less dependent on serendipity? What if this helps funders to make stronger cases that investing into communities and networks is impactful? What if this makes a stronger case for potential members or for members to contribute financially? That’s the opportunity.

Building on this hypothesis, here are five opportunity areas that stand out to me, each deserving further exploration and experimentation:

I - Make community knowledge valuable

Many communities have oooodles of call recordings, notes, chats, Miro boards, photos from flip-charts from past gatherings, and online forums full of people reflecting on the community’s central question. If a community is dedicated to x, there is a great chance that there is huuuge amounts of knowledge, experience and even wisdom about x somewhere in that community. But most of that content is hidden in a dusty corner of the internet. Almost nobody watches old recordings or reads old notes. In a time where nobody has time for anything, those forms of content are simply not user-friendly and therefore not valuable. That’s where AI comes in.

AI can help a community to:

• Capture all the data in one place and make it searchable and interactive.

• It can package that content up in infinitely different ways.

• Some of these ways will turn out to be useful and valuable for the group.

In Marc’s example, this shows up as:

• A GPT that has access to all the community’s content and that you can interact with through an interface within their community, just like you would interact with ChatGPT.

• An automated flow that turns session notes and recordings automatically into short AI-generated podcasts (with automated show notes and thumbnails), which then gets automatically uploaded to popular podcast players like Apple Music, minutes after the session ended. This allows people who weren’t at the sessions to get audio “executive summaries”.

• An AI generated “daily spark” that brings a nugget of practical insight from the archives and shares it automatically in their private discord channel.

This is valuable for several reasons:

• Assuming the group will still have serendipitous encounters of relational learning (through virtual sessions and in-person gatherings), now AI opens up a whole additional universe of asynchronous learning. This allows the content to become valuable for people who weren’t at those sessions, who are not part of the community or who want to come back to that content later.

• For groups that center around learning and peer exchange (such as most Communities of Practice), this allows a group to showcase the value of its collective knowledge to outsiders and make a compelling case for joining or funding the group. For example, if you lead a community focused on immunization logistics and have 7965 pieces of content on the topic, you could enable non-members to interact with your group's collective wisdom through your public-facing website and the intuitive interface of a GPT that seamlessly works across languages and formats. This approach supports peers throughout the industry while serving as a powerful advertisement for membership. After all, while learning from a GPT is useful, nothing compares to the relational learning that happens in the supportive environment of a community.

II - Help a community make latent connections

I see three categories for potential connections in a community: obvious connections, serendipitous connections and non-obvious connections.

Obvious connections are just what they sound like. You work in the same field, have the same passion, live in the same place. You meet because there is a sub-group dedicated to your interest or you pro-actively reach out to someone to exchange notes. Easy to identify, easy to act on.

Serendipitous connections happen by chance. You sit next to each other at a dinner and hit it off. You happen to be in the same Zoom break-out room and have a great 10 min conversation. You bump into each other at a networking event. These encounters can be equally valuable, but they are unpredictable (even though one can design for more serendipity).

Non-obvious connections are all the encounters that could be valuable, but aren’t happening, because you don’t know about them. My sense is that the potential of non-obvious, but valuable connections in most groups is huuuuge. You might unknowingly face a similar challenge. You might unknowingly have faced similar challenges in the past. You might unknowingly have very similar values. You might unknowingly be able to help person x with their request because you know y.

Very often we don’t know that we were supposed to connect with someone until we spent an hour sitting next to them. AI can help with that. It can understand our profiles and pro-actively suggest non-obvious connections. It can matchmake.

Technology has already attempted to do that, either to help with serendipitous connections through randomized introductions, or by using spreadsheets where people describe themselves. These spreadsheets are a lot of work to create, a lot of work to keep up and very limited in use.

Imagine an AI that knows fully who you are, what your challenges, interests and needs are. In Marc’s case, their Spark AI bot combines the private member profile data and their LinkedIn profile data to find relevant matches.

This will quickly bring up valid privacy concerns that will need to be carefully debated. Do we give the AI access to only publicly available information, like LinkedIn profile, blog posts, videos, etc, or are we ok for the AI to read everyone’s emails? Or will this be an opt-in opportunity that some members choose to engage with while others don’t? In Marc’s case, they were - unsurprisingly - very intentional about the process and they co-developed extensive guidelines for how the community wants to engage with AI. This seems key. Using AI in community is not just about the outcomes, but equally about the process of how the community designs it.

III - Better match needs and offers

So much community activity is essentially about search (and why I call communities “human search engines”). People use the relational field of the community to search for relevant answers to their questions. The community - if there is a web of trust - allows for better search results than if you searched publicly. The answers are more contextual. And the answers are more generous: People offer things they wouldn’t offer to a stranger.

However, most of the matching between needs and offers still happens through virtual forums or groups that are organized chronologically. Success depends on whether the person with a relevant offer happens to see a post from someone with a matching need. Effective matchmaking relies on both parties reading these messages at the right time.

Additionally, needs and offers aren't static—yesterday's need might not be relevant today. This is why static approaches like spreadsheets become outdated almost immediately and rarely function beyond an initial burst of connections.

AI can—in theory—understand community members' needs as well as their gifts, assets, offers, and talents. It can then proactively suggest appropriate matches.

Among existing technologies, SuperHive performs well in this area, and we've successfully implemented it across various groups. It makes asking and helping both dynamic and recurring, moving interactions from virtual forums to your inbox. Yet even this system depends on someone actively formulating their need and someone else reading that email within a useful timeframe.

What if AI could understand your needs and continuously, proactively search for relevant matches? What if it helped you recognize your own assets and gifts, then suggested opportunities to help fellow members?

IV - Automate membership processes

Dave Bayless recently pointed out the potential for automation to support community work, especially the sometimes thankless “dirty dishes” of community admin. With automation tools like Make or N8N that incorporate AI, there are now significant opportunities to streamline these processes.

For example, the onboarding process can be greatly simplified: An AI agent can automatically add new members to social networks or WhatsApp groups, update your website's member directory, email upcoming gatherings to new members, and manage yearly membership subscription renewals.

It’s crucial that AI doesn’t replace human interactions, but rather enhances them. Ideally, this means community weavers can focus their energy on meaningful 1:1 calls and creating warm, personal welcomes for new members, while administrative tasks handle themselves in the background. As Engin Ayaz and Melissa Clissold write in this thoughtful reflection on when to incorporate AI automation and when to avoid it: “…let AI handle the repeatable. Let humans hold the irreducible”.

V - Track Key Metrics

Coming back to the original hypothesis, if communities and networks already deliver value but struggle to prove it because much of that value happens in hidden, unexpected ways through 1:1 interactions, AI could help quantify this impact.

• What if we could measure what knowledge exists within our community and how valuable it is for members and beyond?

• What if we could track how many connections form within the community?

• What if we could measure both the quantity and quality of needs matched with offers?

• What if we could quantify seemingly intangible qualities such as trust, care, and generosity?

• What if we could automatically visualize connections?

Some communities currently do this currently. Marc has been surveying his community every 6 months for 4 years, which has given him a profound understanding of the impact they are creating. What if this could be automated going forward?

Together Institute AI Lab: An invitation to experiment

As you probably can tell from the post above, at Together Institute we are really excited to experiment how AI might serve communities in meaningful ways. That’s why we are bringing together some close partners and are putting together an AI Lab. We have a few more spaces open. So if you interested in implementing AI experiments in your community or network and have some budget to invest into the development of simple technology, please reach out (fabian@together-institute.org), we’d love to weave you in.

Till next time, Fab

---

About Entangled

// Entangled: Welcome to Entangled, celebrating the people, insights and practices that help purpose-driven communities & networks thrive. If you care about bringing people together, this might be for you.

// Hello, I’m Fabian, co-founder of the Together Institute where we work with purpose-driven communities, networks and their leaders to help them thrive.

Credits: Thank you to Marc Fonteijn for sharing his experiments and giving comments on a draft of this post. And thanks to my colleague Michel Bachmann for edits. Photo by Skyler Ewing