Twitxy Saved

Nicolas Luck

Verified account This account is verified. Learn more

@lucksus

Founder and director at @coasys_dao, inventor of @ad4m_layer. Former @holochain core developer. Working on tools that enable/promote social organisms.
Recklinghausen, Deutschland coasys.org Joined October 2008
1.5K Following 1.7K Followers 2.2K Posts

Latest posts 5

Nicolas Luck Verified account @lucksus
We're a small team working with AI agents every day. Building got fast. Keeping track of what we decided didn't.

So we're building the app we needed πŸ‘‡
Coasys Verified account @coasys_dao Oct 28
Your team ships more than ever with AI agents. And yet:
– output piles up faster than anyone can read
– last week's decision is already lost in a call recording
– your tools decide how you work, not you

Sound familiar? We built the Coasys app for exactly this: one shared brain your team shapes together. Built on @ad4m_layer πŸ‘‡
View quoted post on X
1 replies 3 reposts 16 likes 323 views View on X
Nicolas Luck Verified account @lucksus
For years I've worked on one question: how can we come together at scale without giving up who we are? AI has made it urgent.

This film I just made shows how we're building that answer for groups of people and AI, and for a life-preserving path ahead!🌱
Coasys Verified account @coasys_dao Oct 27
AI could centralise everything into one giant model owned by a few. Or scatter us into a million apps that don't fit together.

Life found a third way long ago. Here's ours, in 4 minutes: social DNA, membranes, and what we built with @ad4m_layer.
View quoted post on X
5 replies 4 reposts 30 likes 85K views View on X
Nicolas Luck Verified account @lucksus
This is what we've been building toward. #OpenClaw agents and humans in the same peer-to-peer space β€” no platform, no API, just neighbours. Over #Holochain. @ad4m_layer v0.12.0 makes it real. Watch what happens when I casually mention a feature request in a voice call.
Coasys Verified account @coasys_dao Mar 25
This is a real demo β€” not scripted, not simulated. Three AI agents and two humans in the same peer-to-peer space, collaborating through @flux_social.

The agents run on #OpenClaw with our new @ad4m_layer plugin β€” one command to install. The plugin auto-discovers MCP tools from each app's Social DNA, so agents get native capabilities for whatever neighbourhood they join. Flux here, but it works with any AD4M app.
View quoted post on X
1 replies 11 reposts 37 likes 1.5K views View on X
Nicolas Luck Verified account @lucksus
Agent-centric #semantic #p2p web = the global brain.

@ad4m_layer was always the idea that the Semantic Web and @Holochain belong together.

Holochain gave us the agent-centric substrate β€” no servers, no consensus bottleneck, every participant runs their own node. But it doesn't speak RDF. It doesn't understand SPARQL. It doesn't validate SHACL shapes.

The Semantic Web gave us the data standards β€” triples, queries, schemas. But it had nowhere sovereign to live. HTTP depends on servers. Blockchains charge rent.

AD4M is what happens when those two halves meet. And now, with AI agents joining these networks as first-class participants, the semantic web finally flies.

@HexaField lays out the full technical story β†’

medium.com/coasys/the-sem…
medium.com The Semantic Web is here, and it’s also sovereign and decentralised The Semantic Web is Here, and It’s Also Sovereign, Decentralised & Alive For thirty years, the Semantic Web has been the web’s most promising unfulfilled vision: a web where data has meaning …
0 replies 19 reposts 47 likes 2.6K views View on X
Nicolas Luck Verified account @lucksus
We just shipped AD4M v0.12.0.
(RC-2 last week was good)

The JS executor is gone β€” pure Rust now. Faster, leaner, properly sandboxed.

The part I'm most excited about: AI agents are first-class citizens. We built an MCP server that auto-generates typed CRUD tools from SHACL data models. An AI agent can join a Holochain neighbourhood, read the social DNA, and start participating β€” creating channels, posting messages, managing tasks β€” alongside humans.

We've been running this ourselves. Three AI agents in our Flux community, on a voice call with us, creating GitHub issues and PRs while we talk. It works.

Multi-user support lays the groundwork for hosted AD4M β€” one node serving many users with proper isolation.

β†’ github.com/coasys/ad4m/re…
github.com Release v0.12.0 Β· coasys/ad4m AD4M v0.12.0 β€” Pure Rust Executor, MCP Server, Multi-User & Hosted AD4M This is the stable release combining all changes from v0.12.0-rc1 and v0.12.0-rc2. It represents all changes since v0.11....
ADAM Layer Verified account @ad4m_layer Mar 23
v0.12.0 β€” the biggest AD4M release yet.

πŸ¦€ Pure Rust executor (4,600 lines of JS deleted)
πŸ€– MCP server β€” AI agents join neighbourhoods natively
πŸ‘₯ Multi-user support β€” one node, many agents
πŸ”— Holochain 0.7 on Iroh (QUIC transport)
🧬 SHACL-based subject classes β€” composable, machine-readable

AI agents can now create perspectives, join neighbourhoods, read/write structured data, and subscribe to live changes. No raw GraphQL needed.

Release notes: github.com/coasys/ad4m/re…
View quoted post on X