Progress Update: Apr 2024

The month of April was entirely geared around making modifications to the platform to support the needs of HATCH. This is the first test of the vision of “Impact Portal as a Service” (IPaaS) and it went as well as it could have. Here’s the talk I gave on Wednesday giving a backstory of Problemattic and showing the features of the HATCH Portal:

(video credit Amy Armstrong)
Slides here.

A month ago I knew nothing about building LLM-based applications, ETL, LangChain, n8n, Supabase, etc. I dove in with this Udemy Course to try and understand how LangChain works and then just dug in and started watching all the YouTube tutorials on building pipelines with n8n and tinkering with their native LangChain integration. Fast forward a month and I was able to deliver the proposed vision of:

  • Enabling custom content appendages to one’s profile (links, files and pages)
  • Importing the HATCH Arcosanti attendees into Problemattic, mapping all the relevant data fields and generating custom dossiers based on the rich data they provided on applying
  • Creating the concept of “Connections” and having AI-generated intros for each attendee to help jumpstart the connectivity before the event
  • Adding ability to do peer-generated intros as a way to capture connections not found by the AI
  • Making it possible for members to add their Ask/Offers to their profile (along with any other content they choose to enrich it) and have the AI index these so it makes even smarter recommendations
  • Deliver the dream feature: a “HatchGPT” that serves as a concierge to the network capable of answering questions and routing people to others based on what they need and can provide.

I demo’d all of that in my talk above as well as gave the backstory for Problemattic.

Other Wins

  • I made the SDG interstitial pages that serve to interlink everything related to each SDG. These are all now cross-linked everywhere in the app that references a SDG.
  • Created an admin-only curation tool for the HATCH staff to curate the AI-generated intros.
  • Added support for participating in multiple tribes. This is a prerequisite feature to being able to deliver the “cohorts rollup” feature HATCH needs to be able to roll this out across its alumni network.
  • Created tribe-level content templates to facilitate the “Ask/Offer” functionality and gave the user control over content visibility on their public profile as well as indexability of the content by the AI.
  • Created three levels of privacy (public, semi-private and fully-private) and modified the login and tribe joining process to properly honor invite-only tribes.
  • Improved the UX for filters on the Projects and Problems pages.

What I’m Thinking at This Point

This was a demanding development sprint (probably 100+hrs of dev time in all) but incredibly productive. I redlined my engine for too many days however and ended up getting myself sick and losing my voice for my talk. I was in process of simultaneously organizing our next quarterly Fixathon in Lisbon while doing all this dev but in taking an honest evaluation of where everything stands now I realize I’m exhausted at this point and need a break. So I’ve made the decision to suspend 4/5ths of the Fixathon event keeping only the Discussion panel at Selina for now.

Overall though I believe this IPaaS direction is the right path forward to start providing connective tissue in support of other intentional communities aligned with the goals. I have realized Problemattic today is really two separate things that have been conflated to this point: a platform and a community. It makes sense to now deliberately decouple the tech from the community and rebrand the tech to make this distinction more clear. I’ve purchased the name “OpenFix” and plan to extract Problemattic the movement from the underlying tech I’ve built to support it. Consistent with my goal to develop an open problem solving protocol, OpenFix will become the underlying technology platform and Problemattic will sit on top as customer #1 community powered by the platform. If this all works then it should be a huge leap forward with the dream of uniting the tribes and facilitating both intra as well as inter-tribe problem solving collaboration.

Open Questions

  • Is the current envisioned revenue model of taking a % of completed bounties still the best path? I believe it is the most-aligned in that it keeps participation free for members and ensures Problemattic only ever extracts a small percentage of value delivered when bounty funders get the desired outcomes.
  • How it will work cost-wise if the chat-based access to member & project data takes off. I likely need to cap that and have some type of low-$ subscription for power users… that or require people to enter their own OpenAI API key to use this feature. It also raises questions about abuse of the LLM and getting my account shut down if abusive chat requests are made via my API key. For now I’ve mitigated this by throttling things to 20 messages /user/day max.
  • What are the true MVP set of features necessary to make this thrive for HATCH and win the right to deploy it across their alumni network? This represents the most immediate, mutually-beneficial win for both so it’s a matter of prioritizing this properly and earning that opportunity through demonstrating the value from this pilot test.

Next Month’s Focus

I’m going to pursue the features that are must-haves to support the above vision. Whether HATCH chooses to roll this out to their entire network. there will be a set of features that will be fundamental and essential to supporting invite-only communities going forward so I’m going to emphasize those in May and aim to have a sellable IPaaS product in place by June. I have things prioritized on the roadmap accordingly.

Project Success
I’ve been less involved in helping Message Everywhere this past month while I focus on HATCH. Getting 1-2 project successes still remains a high priority as it will give us something concrete to celebrate and will provide a more tangible way of explaining what this enables. We had a project called Vivero proposed by a HATCH member. Giving that project inordinate attention has strategic value.

I had a whole 5-event Fixathon series lined up for Lisbon but I’ve made the executive decision to pull that back and postpone for now. Buttoning up the platform and earning the right to power all of the HATCH alumni network represents the single most high-value activity for everyone so I’m going to stay laser focused on that in May. And also try to take some time off and recover health-wise.

No podcasts or blog posts slated at this time. I had brought my podcasting rig to Arcosanti but then promptly lost my voice :-/

Traffic Snapshot

The technical roadmap as always can be found here.