We're proud to announce that we've now entered a second hackathon: Solana Frontier Hackathon.
The submission will be reviewed and completed towards the end of the month (when Hugr is ready).
First stream back: last minute touch ups and preps for OpenStudio app (you can view the VOD on pump)
Got the payment processor account approved today so should be on track to release to the public tomorrow.
---
New month and you know what that means: I start consistency working on the Hugr web app, I can't tell you the exact timeline but I can outline what to expect over the next couple weeks/month and can be watched live on pump: Hugr landing page. Planning, designing, completing the web app. Smaller feature updates for Rune code, OpenStudio and Hugr framework.
No time to watch the VODS? No problem, I'll write daily updates on here so you're always in the loop.
Btw, I'm thinking about creating a $RUNE community section on the main Rune website. Login with your wallet (with a trusted platform like Privy) and depending on how much of the token you own, get access to chats and channels for organized chatting, voting etc. I think it would be a great place to keep things a bit more organized and cut out the slop. Almost like a Discord channel but exclusively for verified token holders. Let me know your thoughts.
Update on the streaming: I've had some issues with my monitor the past week (sizing issues with mac dock, iykyk). I've ordered a new dock so waiting on that, bear with me. It hasn't affected my work output but just no stream until the new dock arrives (few days I think). I could stream with just a single screen but don't want to risk leaking sensitive info.
Dropping a screen studio alternative soon. Open source. Fully native macOS.
I can't justify paying for a tool that your grandma could vibe code. There are other solutions but they're either paid or horrible to use (yes I'm talking about you "recordly")
@runedotgl is pushing further into AI infrastructure with Hugr, an open source framework for building, chaining and scheduling AI agent workflows through a clean visual dashboard.
It enables custom AI workers, flexible LLM integration and real automation through triggers, scheduling and structured multi agent collaboration.
With built in workflow templates and an AI assistant that can generate agents and pipelines, it moves beyond simple tools toward a full operating layer for AI driven systems.
Notably, the focus is on building real infrastructure rather than directly promoting a token, which adds another layer of credibility.
If execution stability and context handling hold up, this could become a serious infrastructure play.
◾️◾️◾️
Introducing Hugr - An open source framework for building, chaining and scheduling agent workflows.
- Build custom AI workers
- Wire in any LLM provider
- Visual dashboard UI for building workflows, managing workers, configuring providers, scheduling and monitoring sessions
- Built-in AI assistant that can create workflows, triggers and workers for you
- 6 built-in agents
- 5 preset utility agents
- Pipeline templates across 6 categories out of the box: Development, Content, Research, Operations, Data and General
- Schedule workflows with cron triggers
- 20+ trigger templates for common automations
- Full git integration with three isolation modes
- Supervised or fully autonomous execution
- Clarification system
- Structured message passing between agents via an append-only JSONL joblog
- Configurable data storage path
- Agents hand off context
- Skill system
- Full activity streaming
- Session persistence and recovery
Use it as a library in your own apps or run it standalone with the dashboard.
MIT licensed, TypeScript
⭐️Star on: https://t.co/NYH6C5Wklx
@runedotgl made a pivotal change from "AI orchestration tool" to "agentic infrastructure for non-technical users." The product thesis changed from competing with frameworks to becoming the platform everyone uses.
Why This Pivot Is Revolutionary:
Most AI tools target developers. Rune is now targeting the 90% of people who want AI to work but don't want to learn how it works.
The problem:
- Business owners need automated workflows
- Non-technical teams want AI workers
- Nobody wants to spend days setting up agents.
The current solutions:
- Claude Code: Requires terminal knowledge
- Cursor: Built for developers
- LangChain: Python programming required
- OpenClaw: Technical setup needed
The gap:
Ready-made AI automation for people who don't code.
What Makes This Different:
- Circuits - Computational Memory
Not flat text storage. Not vector search. Not markdown files.
A probabilistic knowledge graph where:
- Every piece of knowledge = node with confidence score
- Every relationship = typed edge with weight
- Memory compounds over sessions (Session 1 vs Session 20 intelligence gap)
Contradictions detected structurally
Staleness tracked through decay functions
Entropy measured (how certain is your understanding?)
Example:
- Session 1: Agent learns basic API structure
- Session 20: Agent knows which modules are fragile, which patterns broke twice and how they were fixed, which dependencies cascade when you change X, which tests always need updating when schema changes
This isn't recall. This is reasoning over accumulated knowledge.
2 Skill Creator System
v0.4.0 just shipped:
- Guided skill creation pipeline
- Six pre-built templates
- Skill session management
- Per-agent skill configuration
- Skills marketplace infrastructure
What this enables:
Non-technical users can create custom AI workers through conversation, not code.
"I need an agent that monitors my competitor's pricing and alerts me when they change" → Skill creator builds it.
3. Pay-Per-Usage Web/Desktop App
Coming next:
No setup. No learning curve. Just:
- Choose automation template
- Configure for your use case
- Pay per task executed
- AI worker runs continuously
This is Zapier for AI agents, except the agents actually understand context.
This is the "iPhone moment" for AI agents.
Before iPhone: Multiple devices (phone, camera, GPS, music player)
After iPhone: One device that does everything
Before $Rune: Multiple AI tools requiring technical knowledge
After $Rune: One platform with ready-made automation for everyone.
✅ Circuits memory = agents that get smarter over time
✅ Skill creator = non-technical users build custom workers
✅ Pay-per-usage = no commitment, just results
✅ MCP integration = works with existing tools
✅ Open source core = developer ecosystem compounds
✅ Governance coming = community ownership
$rune @will8bit
Ca: 9fAT1fVVEKMEL6rUeYxSYU8ZjQ5dSiTzAZYTdiPvpump