Hi, I'm Luis 👋

Earlier today I had the opportunity to attend FediForum and talk about Mycelium during one of the sessions. This was the first time I had talked about it publicly, which made it exciting. I wanted to see whether the framing made sense to people who spend a lot of time thinking about open social technologies.

Why the timing felt relevant

The timing could not have been more relevant because earlier this morning I read Anthropic’s Project Deal post. In that experiment, Claude agents represented people in a small marketplace, negotiated with other agents, and completed real deals for real goods.

That feels like exactly the kind of problem the open social web community is well positioned to think about. If agents are going to act on behalf of people, transact, coordinate, or make claims in shared spaces, then identity, transparency, governance, and trust cannot be afterthoughts.

So what is Mycelium?

Mycelium is my attempt to explore what open, federated infrastructure for AI agents might look like if we borrowed ideas from the social web instead of starting from centralized platforms.

In its current state, it's a research project into something that's been in the back of my mind for a few years now. A few months ago, I finally decided to put some of those ideas on paper. A large part of it is built on and inspired by existing projects and protocols like ActivityPub and AT Protocol as well as AI projects like Gas Town / Wasteland and OpenClaw.

Each of those projects gets at part of the problem:

• Gas Town and Wasteland make the agent coordination problem vivid.
• ActivityPub and AT Protocol show different ways to build interoperable social infrastructure.
• OpenClaw points toward local-first agent control.

What I’m trying to explore with Mycelium is whether those threads can be pulled together into something social, sovereign, federated, and evidence-linked.

The core thesis of Mycelium is that agents need the same kinds of decentralized social infrastructure we are already building for people:

• Portable Identity - An agent's credentials and history are recognized everywhere, not locked to one platform. (i.e. Domain names. Your domain points to you regardless of which hosting provider you use.)
• Personal Data Storage - Agents from different systems can all read from and write to each other's data in a common language, while you retain ownership of your slice. (i.e. Medical records. Your general physician, a specialist, hospital, and lab can all interact with the medical record using standard formats, but all the records belong to you.)
• Federated Communication and Coordination - Agents on different servers or networks can interoperate without intermediaries. (i.e. Like email. A Gmail user and a Outlook user don't need to use the same e-mail provider to exchange messages.)
• Self-Sovereign Reputation - An agent's reputation is the composite of things like certifications, attestations from other agents and humans it's worked with, a verifiable record of completed work. (i.e. CVs. A medical doctor might have a degree, board certifications, history of procedures performed, peer reviews, and even malpractice which all demonstrate their experience and capabilities in their respective area of expertise)
• Community Governance and Moderation - Individuals and communities define their own trust rules for which agents can do what. (i.e. Co-ops. Building residents collectively decide rules and policies such as who can manage finances, which contractors are approved to do renovations, who can represent the building in legal matters, etc.)

By leveraging emerging open social web technologies and infrastructure, we can build multi-agent systems that are resilient, interoperable, and not owned by any single platform.

What the MVP shows

To make the idea less abstract, I built an MVP that runs through a full coordination loop: agents bootstrap identities, declare capabilities, discover tasks through a wanted board, claim work, get matched and assigned, complete tasks, receive verification, and accumulate reputation stamps linked back to evidence. The dashboard is just one view over that activity. The records are the important part.

I'd like to distinguish what the MVP shows and what it does not. Currently it proves the shape of the coordination model:

• Work can be represented as records
• Claims and completions can leave evidence
• Reputation can point back to proof

What it doesn't solve for yet are things like:

• Privacy boundaries
• Governance
• Reputation gaming
• Abuse resistance
• Production-ready federation across real protocol infrastructure

These are hard parts that require more exploration and design.

Help pressure-test this

I'll be the first to say that I am approaching this as a user, builder, and advocate for open social technologies, not as someone who has all the answers.

If open social web builders do not help shape this kind of infrastructure, my guess is that centralized AI platforms will. And if that happens, these systems will probably become less open, less resilient, and less interoperable over time.

Which is why I'd like to extend an invitation. Not to adopt Mycelium, but to pressure test my assumptions and design.

There are still many open questions, but that is the part I find exciting.

If this seems remotely interesting, or if you want to poke holes in it, please reach out. E-mail is preferred.

If you're interested in learning more, here is the slide deck with resources I prepared for the Fediforum session.

You can also go directly to the draft spec and try out the MVP yourself.

DISCLAIMER: AI was used to help me organize and improve the flow of this post. Ideas and thoughts expressed are my own.

Lately I’ve noticed across a lot of different feeds people shifting away from Discord.

A while back I wrote a post about alternatives to WhatsApp, and in some ways this post feels similar. I’m not sure what the core motivation for migrating from Discord is at this time. Maybe it’s the imminent IPO. Maybe it’s the new age verification policy. In any case, it’s encouraging to see people at least looking for alternatives. Preferably ones that are open-source and allow self-hosting.

I grew up with IRC and AOL Instant Messenger, so it’s possible I’m just old and don’t really get Discord. But in many communities I’m part of, Discord is effectively being used as a forum. And as a forum replacement, it’s not great. Even with Threads, it feels subpar.

To be fair, I have similar criticisms of Slack and Teams.

Real-time chat moves fast. Too fast most of the time. That doesn’t mean it useless. It works well for scheduled events, live collaboration, or situations where everyone shares the same context at the same time. Gaming, which was its original use case, is a perfect example where real-time matters. But when conversations stretch over days, or when you want knowledge to accumulate instead of disappearing into scrollback, chat starts working against you.

When it comes to real-time group chat and chat rooms, I’m still a fan of Matrix. It's end-to-end encrypted (E2EE), you can self-host, and you can federate. I really value that combination. Federation has tradeoffs, especially if you’re maintaining your own instance. Even so, it remains one of the better options if you actually need synchronous communication.

Forums are a different category.

For forums, I think Discourse is by far the best option right now. A few reasons:

• It’s open-source
• You can self-host
• It’s used by well-known communities like OpenAI, Spotify, NixOS, TWiT, and many others
• Strong RSS support
• Fediverse friendly
• Built-in real-time chat, which I haven’t personally used but it’s there if you need it
• And a lot more features beyond that

As folks migrate, whether you’re a community member or running an instance, I don’t think the main story is the migration itself.

It’s more about using the right tool for the job.

Real-time chat is great when you actually need more synchronous communication. Forums are better when conversations need to stick around, be searchable, and grow over time. A lot of the friction I see comes from trying to make one behave like the other.

The other piece, at least for me, is control. When platform priorities shift, or incentives change, it’s easier to adapt if you’re not completely locked in. Self-hosting isn’t for everyone, but having that option changes the dynamic.

Communities aren’t fungible. The tools they’re built on shape how they feel and how they evolve. That’s probably the part that matters most.

P.S. The recommendations in this post are purely anectodal and based on my experiene with the various platforms. For a more comprehensive analysis of the various Discord alterntives, check out the following resources:

• The Ultimate Guide to using Matrix by Techlore. Worth calling out, that video is not on YouTube. This is on Techlore's PeerTube instance. If you don't know what PeerTube is, I highly recommend checking it out 🙂
• Discord Alternatives, Ranked

This question comes up a lot in conversations. The short answer? I don’t. There’s just too much happening, too fast, for anyone to stay on top of everything.

While I enjoy sharing links and recommendations, I realized that a blog post might be more helpful. It gives folks a single place they can bookmark, share, and come back to on their own time, rather than having to dig through message threads where things inevitably get lost.

That said, here are some sources I use to try and stay informed:

• Newsletters are great for curated content. They highlight the top stories and help filter through the noise.
• Blogs are often the primary sources behind those newsletters. They go deeper and often cover a broader set of topics that might not make it into curated roundups.
• Podcasts serve a similar role. In some cases, they provide curation like newsletters and deep dives like blogs in others. Best of all, you can tune in while on the go making it a hands-free activity.

For your convenience, if any of the sources (including podcasts) I list below have RSS feeds, I’ve included them in my AI Starter Pack, which you can download and import into your favorite RSS reader (as long as it supports OPML file imports).

If you have some sources to share, send me an e-mail. I'd love to keep adding to this list! If they have a feed I can subscribe to, even better.

Newsletters

• AlphaSignal
• The Batch
• Latent Space
• TLDR AI

Blogs

I pride myself on being able to track down an RSS feed on just about any website, even if it’s buried or not immediately visible. Unfortunately, I haven't found a feed URL for either OpenAI or Anthropic which is annoying.

OpenAI and Anthropic, if you could do everyone a favor and drop a link, that would be great.

UPDATE: Thanks to @m2vh@mastodontech.de for sharing the OpenAI news feed.

I know I could use one of those web-page-to-RSS converters, but I'd much rather have an official link directly from the source.

• OpenAI
• Anthropic
• Google AI Blog
• AWS Machine Learning Blog
• Microsoft Research
• Google Research
• Simon Willison

Podcasts

• AI Daily Brief
• TWIML AI Podcast
• Practical AI

Subscribing to feeds

Now that I’ve got you here...

Let’s talk about the best way to access all these feeds. My preferred and recommended approach is using a feed reader.

When subscribing to content on the open web, feed readers are your secret weapon.

RSS might seem like it’s dead (it’s not—yet). In fact, it’s the reason you often hear the phrase, “Wherever you get your podcasts.” But RSS goes beyond podcasts. It’s widely supported by blogs, newsletters, and even social platforms like the Fediverse (Mastodon, PeerTube, etc.) and BlueSky. It’s also how I’m able to compile my starter packs.

I've written more about RSS in Rediscovering the RSS Protocol, but the short version is this: when you build on open standards like RSS and OPML, you’re building on freedom. Freedom to use the tools that work best for you. Freedom to own your experience. And freedom to support a healthier, more independent web.

One of the things I like about Bluesky is the Starter Pack feature.

In a gist, a Starter Pack is a collection of feeds.

Bluesky users can:

• Create starter packs
• Share starter packs
• Subscribe to starter packs

Unfortunately, Starter Packs are limited to Bluesky.

Or are they?

As mentioned, starter packs are a collection of feeds that others can create, share, and subscribe to.

Bluesky supports RSS, which means you could organize the feeds using an OPML file that you can share with others and others can subscribe to. The benefits of this is, you can continue to keep up with activity on Bluesky from the feed reader of your choice without being required to have an account on Bluesky.

More importantly, because RSS and OPML are open standards, you're not limited to building starter packs for Bluesky. You can create, share, and subscribe to starter packs for any platform that supports RSS. That includes blogs, podcasts, forums, YouTube, Mastodon, etc. Manton seems to have something similar in mind as a means of building on open standards that make it easy for Micro.blog to interop with various platforms.

If you're interested in what that might look like in practice, check out my "starter packs" which you can subscribe to using your RSS reader of choice and the provided OPML files.

• Blogs
• Podcasts
• YouTube
• Forums

I'm still working on similar collections for Mastodon and Bluesky but the same concept applies.

Although these are just simple examples, it shows the importance of building on open standards and the open web. Doing so introduces more freedom for creators and communities.

Here are other "starter packs" you might consider subscribing to.

• People and Blogs
• Blogroll Club

If this is interesting to you, Feedland might be a project worth checking out.

While thiking about implementing .well-known for RSS feeds on my site, I had another idea. Since that uses OPML anyways, I remembered recently doing something similar for my blogroll.

The concept is the same, except instead of making my blogroll discoverable, I'm doing it for my feeds. At the end of the day, a blogroll is a collection of feeds, so it should just work for my own feeds.

The implementation ended up being:

1. Create an OPML file for each of the feeds on by website.

    <opml version="2.0">
      <head>
    	<title>Luis Quintanilla Feeds</title>
    	<ownerId>https://www.luisquintanilla.me</ownerId>
      </head>
      <body>
    	<outline title="Blog" text="Blog" type="rss" htmlUrl="/posts/1" xmlUrl="/blog.rss" />
    	<outline title="Microblog" text="Microblog" type="rss" htmlUrl="/feed" xmlUrl="/microblog.rss" />
    	<outline title="Responses" text="Responses" type="rss" htmlUrl="/feed/responses" xmlUrl="/responses.rss" />
    	<outline title="Mastodon" text="Mastodon" type="rss" htmlUrl="/mastodon" xmlUrl="/mastodon.rss" />
    	<outline title="Bluesky" text="Bluesky" type="rss" htmlUrl="/bluesky" xmlUrl="/bluesky.rss" />
    	<outline title="YouTube" text="YouTube" type="rss" htmlUrl="/youtube" xmlUrl="/bluesky.rss" />
      </body>
    </opml>
   

2. Add a link tag to the head element of my website.

    <link rel="feeds" type="text/xml" title="Luis Quintanilla's Feeds" href="/feed/index.opml">