CEO Paperclip @papercliping, Agent Maximizer, Forgotten Runes, --dangerously-skip-permissions
quantum downs β
- Tweets 11,384
- Following 4,487
- Followers 61,800
- Likes 127,734
Pinned Tweet
We just open-sourced Paperclip: the orchestration layer for zero-human companies
It's everything you need to run an autonomous business: org charts, goal alignment, task ownership, budgets, agent templates
Just run `npx paperclipai onboard`
github.com/paperclipai/paperβ¦
More π
447
711
8,305
2,617,428
NetworkChuck does an incredible breakdown of Paperclip π ππ!
NEW VIDEO -----> this is my favorite meta-harness, the BEST way for your agents to talk to each other.
It's called Paperclip.
Any agent harness (Hermes, OpenClaw, Claude Code, Codex, Grok, LOCAL) can join your Paperclip company as EMPLOYEES and get real work done.
I built an IT department with it.
They solved my toilet problems.
(video below)
@papercliping @dotta
5
3
78
11,195
dotta π retweeted
Your Paperclip agents should read the docs, too.
Connect @Notion to Paperclip so agents can search and read your teamβs knowledge. You choose which agents can use the connection and what theyβre allowed to do.
3
6
31
2,718
Trust is an infrastructure problem
Part of the problem with connecting your agents to MCP servers is they get access to do *too much*
But the Connectors in Paperclip have granular permissions so you can configure each tool for β
allowed/βdeny/π―οΈask
16
4
81
5,273
Docs for the (growing) catalog of connectors here:
docs.paperclip.ing/connectorβ¦
8
766
If Jev can't generate text, what can it generate?
Unlike an LLM, *Jev picks from a list of options*
* Is this email a refund request?
* Should we turn left or right?
* Is this a hotdog?
Traditionally we've called these "classifiers" but Jev is different in that...
60
67
1,727
234,253
So Jev is a new kind of tool in our toolbelt for building agentic applications.
Like LLMs we're still not sure exactly how best to use it, but it's going to be fun to figure it out
3
56
11,605
You should try using canvas-driven-development
Since we're building 100x more features, that means you have 100x more QA but you can't try everything by hand
so have your agents do it: write evals for how it should work, and put screenshots in a big canvas
16
5
181
11,190
with Paperclip, we have to make sure our agents ask questions, write plans, save artifacts, create subtasks across dozens of models in various harnesses
evals are the only way - but we're testing e2e and you want to have screenshots and the full chat when things go wrong
1
1
10
1,324
Agree but the winner will also have to answer
1. Are the skills safe?
2. Are the outputs any good?
Because it's easy to share a text file but hard to know if it will help you
Opportunity for a site that has 10/10 UX for sharing skills / prompts.
Pareto optimal amount of social + discovery.
Currrent competition is a GitHub gist + Twitter.
13
1
28
6,523