Fusion ⚫️

The fundamental problem I’m trying to solve with Fusion - I want tasks, agents (and agent companies), missions, and everything in my dev environment (git, files, worktrees, etc) working with any model (local or cloud), on any device (laptop, Mac mini/studio, Linux server, cloud VM, phone), all accessible from anywhere and with shared state across each machine, session, agent. I want it all working together with a single pane of glass. Fusion. Coming very soon. Open Source. Built with love (and with @dotta’s @Paperclip_AI and @NousResearch @Teknium Hermes agent, leveraging @badlogicgames’ pi as an engine). Standing on the shoulders of giants…

@dotta @jamesbrooksco Same

@jamesbrooksco It already has a nice mobile web ui - I use tailscale to use Paperclip while on the toilet

Really hope we see a native mobile app for @dotta’s paperclip. dreaming of getting tickets assigned to me and making comments on the toilet

@naveen4 @mercury_build @dotta Curious the difference in target audience?

@runfusion @mercury_build @dotta running on the cloud is actually a very intentional feature for us - Paperclip and Mercury have different target audiences

Introducing Mercury. Today, we have two types of “AI teams”: 1. Agents in tabs, terminals, 12 browser windows 2. One workspace. Humans and agents actually coordinated. We built @mercury_build so everyone can be type 2. Drop who you are or what you're working on — we'll build your org and send you your agent team.

@JamesNguyen868 Running local but online models

@runfusion online or local?

112 agents running

@naveen4 @mercury_build Feels like a non starter when Paperclip exists and is free/open. @dotta

@runfusion @mercury_build it isn't, we're a cloud based application. however, you can connect local agents (ie. claude code, openclaw) into mercury!

Need to fix spacing of the icons on the buttons

@LuciferMornens @_imdawon Open source, releasing this week

@runfusion @_imdawon Care to share? or is it closed?

its truly remarkable how bad most models are at coding GPT 5.4 xhigh and Opus 4.6 are in a league of their own been using lots of Chinese LLMs this weekend and uhhhhh wow they are not very good

@onlinedopamine Use Hermes

what model are you guys currently using with openclaw? i switched to codex after anthropic cut of access it's been absolutely horrible so far all of my cron jobs broke, tool calling seems non existent, and it's horrendous at following instructions PLS SEND HELP

@andreneves Amazing will let you know asap dm me

@runfusion Happy to beta test and feed things upstream whenever the time is right.

Pi has a great composable ecosystem. Fusion is built on Pi and can leverage all of these underlying extensions natively

@appfactory Fusion should run in cloudflare once it’s launched

If I don't get a Pi agent harness running in cloudflare I'm going to be sad.

@LuciferMornens @_imdawon Not a CLI - it’s fusion a multi agent orchestrator

@runfusion @_imdawon What CLI is that?

@appfactory This is close to what Fusion is doing

Pi harness + workers + virtual fs + codemode + glm 5.1

@aronprins @dotta @Paperclip_AI Will be open sourced soon - this is what I pinged you about on discord

@runfusion @dotta @Paperclip_AI Looks good! PR? Is this already worked on by someone else?

So uh @dotta I think Fusion is close to having fully implemented the agentcompanies.io spec - including import from companies.sh. The first end to end Paperclip-compatible agent companies harness (other than @Paperclip_AI)? cc: @aronprins

@MatthewBerman @NousResearch @Teknium Yes for coding agents. It can run as many parallel agents as your hardware can handle. Would love help and contributions! Let’s build!

@runfusion @NousResearch @Teknium Is this for coding agents? I want to build something like this but for maximum parallelization of coding agents (I don’t want to build the harness)

Weekends are for hacking! What are you building with Hermes?

Soon.