Update: Scaling Back The Foundation
I launched The Foundation with big plans: federated AI knowledge commons, MCP servers, semantic search across organizations. Built the proof-of-concept. It works.
But I underestimated the scope. This is a multi-person, multi-month project. And I have client work that pays bills and family that needs time.
What Changes
Right now:
The Foundation becomes simpler - a place where developers write verification case studies. No complex infrastructure. Just knowledge sharing:
- "AI code I rejected and why"
- "How I verified this architectural decision"
- "Times AI was confidently wrong"
Later (maybe):
The federated infrastructure. When we have people and capacity to build it properly.
If You Want to Help
- Write verification case studies
- Share examples of rejecting AI output
- If you're technical and want to co-build infrastructure: DM me
I'd rather be honest about capacity than burn out delivering promises I can't keep.
The problem is real. The solution is still needed. Just needs to grow at a sustainable pace.
Top comments (19)
Hey @richardpascoe and @dannwaneri! Hope you are well. Apologies for my late response. I saw your comments during the weekend, but was not able to respond at that time.
I need to see more on "how I verified this decision", but one thing on the top of my mind is can it reproduce the same result if I were to put this info in? It's brute force, but that is one of the solution I can think of based on my understanding. I have to reread your guy's key posts in order for me to fully grasp.
Additionally, I took a look at the repo and will try to set up my environment in a bit. A lot of things that needs my attention at the moment. My apologies to you both! If there is anything you guys would like for me to do, let me know!
Yeah completely understood - building the POC and actually building out/running the infrastructure "at scale" are two different things - it's not that one is more "difficult" than the other, it's more that it's gonna be time consuming ...
And what about the costs of running all that infrastructure? (including AI tokens being consumed ...)
Good (great) that you're honestly signaling this early on, and that you're setting your priorities the way they should be!
@leob Thanks for the thoughtful comment.you nailed it exactly.
The POC was fun and validating (got the core import/security/embeddings/federation basics working in weeks), but yeah turning that into reliable, always-on infra at any real scale is a completely different beast. Time sink is one thing; the ongoing costs (D1 reads/writes, Vectorize indexing, Workers AI embeddings, potential bandwidth if federation picks up) add up fast even at modest usage. I didn't want to quietly rack up bills or burn out trying to optimize everything solo.
Super grateful you're calling it out. transparency early feels way better than pretending it's all smooth sailing. Appreciate you getting it.
Being transparent about these things is so valuable, it's the right thing to do - so many things go wrong just because people want to keep up the pretense, and want to keep unrealistic expectations alive ... and you clearly did your part, it's for others to step in and contribute if they agree this is an important project!
@richardpascoe Thanks for the update. this is exactly the kind of creative, grounded thinking I love seeing in the community.
TiddlyWiki + TiddlyServer + Git/GitHub as a middle-ground sounds super pragmatic. No exotic protocols, built-in version history, easy redundancy via clones/forks, and public accessibility through GitHub Pages.that's a solid foundation for distributed knowledge without the full federation overhead. I've seen folks use TidGi-Desktop or GitHub Saver for auto-backup/sync, and scheduled pull/push (cron or Actions) keeps it low-maintenance.
Love that you're exploring this while we test the federated side slowly. No rush. family and client work come first for all of us. If you get a small POC going (even just a repo with auto-sync demo), share a link or quick write-up whenever you're ready. I'd be happy to link it from The Foundation tag or try importing some Claude captures as tiddlers to see how it flows.
@francistrdev — if you're around, curious if this resonates with any ideas you have brewing. Either way, grateful for the momentum while we keep things sustainable.
Onwards.one thoughtful step at a time.
@richardpascoe Thank you. that means a lot coming from you.
Seeing you publish under The Foundation tag and actually engage with the early stuff has been one of the best parts so far. You're right: family first is non-negotiable, and client work keeps the lights on. Scaling back isn't quitting. it's making sure this thing can actually survive long enough to matter.
This pause gives breathing room for everyone (me included) to catch up, think clearly, and maybe contribute in smaller, sustainable ways. If you or anyone has a quick "AI code I rejected and why" or "how I verified this decision" story sitting around, I'd love to feature or co-post it when I'm ready to drop the first manual case studies. No pressure, just throwing it out there.
Grateful for the support.let's build something special at a pace that doesn't break anyone.
@richardpascoe @dannwaneri Hey! I made an issue on the Repo about accessibility and it may solve the HTML issue since users maybe using ChatGPT or other AI chats. github.com/dannwaneri/chat-knowled...
Let me know!
@francistrdev Thanks for opening the issue.that's super helpful. Yeah, the HTML import is super Claude-specific and brittle, so making it more accessible (and opening up to ChatGPT/other AIs) would be a big win.
I'll check the issue now and comment there. No rush on env setup or anything.take your time.
@richardpascoe This could pair nicely with your TiddlyWiki/Just-the-Docs ideas too.
Appreciate both of you keeping things moving. I'll reply in the issue soon.
Sounds good! To confirm we are aligned, can you send me the links to the articles that you think it is important? Just to make sure we are reading the same thing
Ok Thanks! Do let me know!
Hey @richardpascoe and @dannwaneri! Hope you are well.
For WIKI related stuff, you can also look into "Just-the-docs" that we used for documenting Summit Events in Salesforce. sfdo-community-sprints.github.io/. Again, something on the top of my head. Do let me know what you think!
Honest pivot. Scaling back to focus on what works shows real maturity.
The trap: Building for hype vs shipping for users.
Respect the decision—most projects die from overambition, not underambition.
Better to nail one thing perfectly than ship ten half-baked features.
What's the core problem you're doubling down on? 🚀