Hello again and happy summer!
We’re back from an extended holiday weekend - June was a packed month and we needed some time off ;)
Cosmik
Check out our long-due writeup of ATScience 2026 (March 2026, Vancouver) with the ATScience crew:
ATProto Science|June 04, 2026
June was also full with exciting science events.
Ronen attended the Foresight Vision Weekend in London to continue the BiTS field building work. Lots of great metascience discussions, about how to wisely integrate AI for science, and the new funding models and institutions this moment calls for.
Following that was the MIRA (Modular Interoperable Research Attribution) workshop in beautiful Ireland - an intense 4-day workshop where researchers, developers and field builders gathered to design and develop a new modular medium for research. Follow our progress at https://www.mira.science/
We also joined a panel discussion on the future of scientific infrastructure organized by Catalyze as part of their “Frontiers in Research” series. Moderated by Sandeep Patel (Catalyze), with (openRxiv), (University of Maryland/Discourse Graphs), and (Cosmik). See the recording and accompanying show notes.
Semble
We launched the Semble API!
Knowing the awesome community here, maybe we shouldn’t be surprised by this anymore, but still - within hours people were hacking on the API (and skipping meetings to do so 🙃) to develop new integrations and experiences! It is so exciting to see the energy! Check out the Community section below for details.
skipped atproto meetup yesterday to stay home and play with new @semble.so API. #Irony #SorryNotSorry
And - we’re excited to see what you build with the Semble API!
Drop us a line at hello@cosmik.network or tag us to share your creations
We made a public board where you can add feature requests (and vote on them!) or ask questions about Semble! Thanks to for creating this excellent tool.
We’re working on a new version of the Semble landing page - more to share soon!
Community
Semble API projects
used the API to create an initial Semble MCP server for AI assistants. You can now talk to your library in ways that would be science fiction just a few years ago (an we have been dreaming about them since the first days of the project!). Some use cases: “Compare my collection X to [handle]'s collection Y. What are they reading that I'm missing?” “Who else saved this link and what did they note about it?”
used the MCP for creating new connections as part of a writing project
is working on a new Semble search engine
👀 gonna drop this soon! been poking around with it for a few hours now! @semble.so thx for the API! #atproto
created a web extension that shows you links you Sembled in orange - lots of exciting potential here to layer in social signals for a co-augmented browsing experience
saw exciting potential for layering in further social context:
Building on some of @chrisshank.com's work, what if we had browsers that were atproto-native? orange link = in your PDS green link = in the PDS of someone you follow Dreaming about the future of social hypertext ✨
...which quickly gained momentum and @me.webbeef.org turned it into a working prototype within a few days 🤯:
networked collective intelligence in action: - @chrisshank.com explores changing the link colour in a browser if saved to @semble.so - I expand it to atproto-wide social context - @nandi.uk asks about atproto browsers - @erlend.sh brings up beaver browser by @me.webbeef.org - prototype this wknd??
has wanted to create a Semble-powered link blog for a while, and launched one this month at https://bmc.bringyourown.computer/links/!
Tab Sanity
“Tab sanity” is interesting emerging theme. We all have so many interesting things to read, so little time. Instead of that menacing and ever-increasing row of open tabs, people are now realizing they can “upcycle” open tabs by saving them to the knowledge commons via Semble. “I may not have time to read this open tab yet, but someone else might benefit from it. And I can finally close that tab!”
used the new Semble API to Claude-code a bookmarks-to-Semble workflow:
Just pushed a bunch of cards to my @semble.so collection via the API from my bookmarks! thank you opus and thank you @cosmik.network team for the API! tab sanity 🤝 contributing to the knowledge commons 🎉
created a Firefox-to-Semble integration
Discord bots
built a Discord–Semble bot that posts every link shared in their Discord to a Semble collection. Ronen followed the collection, found great links, and joined the Discord for the conversations around them. A promising pattern: labs that live in Discord/Slack can pipe their link streams out to the ATmosphere, sharing their curation with everyone outside the server.
@burrito.space‘s Discord-Semble integration (disjecta.portable.agency) just triggered an aha moment for me 💡 They have a bot in their Discord that posts all links shared there to a @semble.so collection. I was following that collection on Semble and found some cool links there, so >
ran with the idea and connected their research lab's link channel, so anyone can follow what CoLab is reading. There's even a dedicated feed at .
Want to see all the links CoLab 🐶 is thinking about? bsky.app/profile/cola...
Atmosphere integrations
launched major improvements including a nice Semble integration showing Semble activity around links on the platform
launched support for syncing saved articles to Semble and Margin 🔄
created a really cool web component that shows how the ATmosphere reacted to any page, including Semble saves. atproto powered altmetrics coming soon? 👀
ATMentions: a web component that pulls how the ATmosphere reacted to any page. Bluesky, @margin.at , @semble.so , @frontpage.fyi , standard.site compatible. Easy to setup: one tag you drop on a page; it surfaces how the ATmosphere reacted to the page — likes, reposts, Margin notes, Semble saves...
uses Semble Connections to help create their "Related Reading" section. Peer-powered curation in action!
We now show sections for: - articles that cite the one you're looking at - articles that have connections defined on @semble.so
Collective Sensemaking in action
spun up an open collection to track the W Social story (a controversial new European Bluesky alternative ICYMI). The collection now has 75 cards from 7 contributors, with 10 followers (and counting!). We think this is an interesting sensemaking pattern that can be applied more broadly, wherever communities want to stay abreast of an event. We see lots of product potential here as well, such as the option to subscribe to a collection to get push notifications, or AI to summarize contributions in real-time.
Real time collective sensemaking in action on @semble.so - I wanted to track the W Social story so I created an open collection. 2 days later it has 27 cards, 7 followers and 7 contributors. And I have a much better handle on what’s going on! >
On our radar
(Follow the “On our radar” collection on Semble)
Creative Reading: Scaffolding Reading for Transformation (Liu et al.) argues most reading-augmentation tools optimize for efficient information extraction ("reading to discard"), and proposes instead scaffolding “reader transformation,” where readers add creative interpretations and are transformed in unexpected ways by the text This research is deeply resonant as Semble is exactly the kind of social hypertext system for scaffolding sensemaking and transformative learning journeys, for both individuals and collectives.
AI for Science Curation (Ben Reinhardt) imagines new curation tools that filter the social-media firehose to surface what actually matters. Importantly, while AI is used for scalable aggregation, these systems ultimately require human taste. Our read - as curation gets recognized as an essential part of the research process, the opportunity is to move or mirror that work onto open social apps like Bluesky and Semble, where AI tools can actually use it, rather than the situation today where data is locked inside closed platforms like LinkedIn or Twitter.
Science as Collective Sensemaking (Charles Yang) makes the case that the substance of science is collective sensemaking, not papers or data: “The truly visionary AI for Science company is not automating experiments or AI-generating Nature papers, but building technology to improve the collective sensemaking ability of scientists.” We appreciate the shoutout to as an example (along with alphaXiv) of a startup trying to build that kind of new sensemaking infrastructure!
Happy hacking 🛠️
Cosmik Team