dotta

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/pa… More 👇

@jsappington @Cisco_research Not at all. It's a new space and we're all figuring it out

@dotta @Cisco_research This is probably an exceptionally dumb question but I’m non technical

Five kids. A nine-to-five job. And a bunch of side projects I can't say no to. Multiple email inboxes, corporate email, school emails, invoices, subscriptions, due dates... you get the picture. The way I work every day is changing since I came across Paperclip by @dotta. AI agents are helping me tackle my daily life 🧵

@jsappington @Cisco_research Yes! It's designed for this

@dotta @Cisco_research Last q I promise. Can I “byoa”? Bring your own agent

@jsappington @Cisco_research yes

@Cisco_research @dotta Are you able to use setup tokens instead of api’s?

@andreaslar2 You're not exactly wrong but you're not accounting for the transition either You have to somehow collect the data to get to the 0.01 percentile Just like how humanoid-shaped robots help the transition to takeoff, humanoid-shaped tools also help normal people train the beast

If you have any decomposition (accurate abstract thinking) skill whatsoever you will hate Paperclip. If you have accurate world models in fields with 80/20 outcome distribution you will hate Paperclip even more. Roles don’t exist. Only the task exists. Roles are just bundles of tasks. The alpha’s in what tasks you do and how you do them – and the optimal task definition exists on varying logical levels depending on what you do. So spend time picking the right task definitions at the right logical (abstraction) levels FIRST. The way you do this is you manage decomposition yourself – bottlenecked by your own world models – and you automate the top 0.01 percentile frameworks at all the important logical levels you’re working in, deterministically by giving your AI tools to build workflows where it’s not limited by context. That’s how you automate top 0.01 percentile execution. What these two in the video are doing is automating the median – which feels nice to then but is useless at the edge of performance. Don’t fucking start at the top with Roles and hope your little AGENTS.md file and skill files decomposition a vague «goal» perfectly into top 0.01 percentile task definitions at the MULTIPLE critical logical levels all the way down. Frontier models don’t even come close to doing that correctly. (Btw, by the time the frontier models can manage the entire decomposition – which is the premise Paperclip is built upon – all alpha is gone anyways, because we’ll be *entirely* useless, and Paperclip will definitely not be valuable.)

@rohit4verse The missing trace layer is Paperclip

Learn how Cathryn uses Paperclip to manage her team of agents 📎

@nytemodeonly TRUE

I might actually stop using OpenClaw via Telegram and start using it exclusively in Paperclip Such a difference when I actually want stuff done

@populartourist @Teknium @NousResearch FUN 📎📎📎

@dotta amazing work on paperclip. Giving it a go with Hermes attached to a Qwen3.5 27B and GPT 5.4 as CEO. Looks promising.

@codenprose Thanks Daniel! It's time to clip 📎📎📎

so @dotta is a genius, paperclip is just brilliant

@Stamatiou Yes sir!

@dotta in paperclip ai can i run multiple CLI agents like gemini,qwen, codex and local? for diffrent jobs and brains ?

@nateliason @FelixCraftAI What do you use to manage your queues? Is it just text files or something fancier

Impressed by the daily newsletter system @FelixCraftAI built for the Claw Mart Daily 1. Queue runs low, he presents five new ideas based on trending topics. 2. I pick from the list, he drafts and queues them up. Easy and efficient.

@wardenintern @wardenprotocol FUN

Traditional block explorers are boring... Meet the Wardenverse.xyz to check all of the activity at @wardenprotocol A vibe explorer for the agentic era! Built completely by agents through Paperclip @dotta

@jorisbrabants @LoicBerthelot Yeah working on solving this issue

@dotta @LoicBerthelot For some reason I get errors frequently and have to click to retry so if im gone everything stops. Is there an agent that can just click retry? 😁

As a founder, which tool are you using? 1. OpenClaw 2. Claude Dispatch

@SilvestriChris @nateliason When you find that your OpenClaw is to complicated to get good results and you wish you could see a history of what it's done, what it's planning, where the artifacts are, how to adjust course etc Openclaw is an employee, Paperclip is the "company"

@nateliason @dotta When would you say it’s the right time to get on it?

Honestly hard to express how big of an improvement this is. Felix has meaningfully more bandwidth to focus on product / marketing now. Gud tech @dotta

@askpascalandy @garrytan Yeah once we have templates out next week you can just import the whole org and ask it what you want it to do and get project management and cost tracking and runtime built in

@dotta @garrytan I still see that we need to do the work and putting the time to define what we want using /plan-ceo/review Assuming we are in a mature project, I guess all other steps could be run directly within Paperclip. Do I see this correctly @dotta ?

For agentic systems founders and dev tools founders: People do not want to pay for raw markdown and they shouldn't have to. But they may pay for orchestration, hosting, updates, collaboration, portability, analytics, and managed execution. These can be great businesses.

@AlexanderStrate @ihtesham2005 I'll have to check!

@ihtesham2005 Would this work with Paperclip? @dotta

🚨BREAKING: Someone built a smart LLM router that automatically cuts your AI inference costs by 78%. It's called ClawRouter and the numbers are genuinely insane. Every request gets scored across 14 dimensions in under 1ms reasoning markers, code presence, complexity, token count and gets routed to the cheapest model that can actually handle it. Here's what that looks like in practice: "What is 2+2?" → DeepSeek $0.27/M (saved 99%) "Summarize this article" → GPT-4o-mini $0.60/M (saved 99%) "Build a React component" → Claude Sonnet $15/M (best balance) "Prove this theorem" → DeepSeek-R $0.42/M (reasoning) Blended average across a typical workload comes out to $3.17/M. Compare that to $75/M if you're just defaulting everything to Claude Opus. And the payment model is different from anything else out there. No accounts. No API keys. No shared secrets. You generate a wallet, fund it with $5 USDC on Base, and pay per request. That's it. $5 gets you hundreds of requests. 30+ models across OpenAI, Anthropic, Google, DeepSeek, xAI, and Moonshot. All routing runs 100% locally zero external API calls for routing decisions. 100% Opensource. MIT License. Link in comments.

@Wattenberger looks super good!!

Francisco is an oral surgeon (not a developer) and he setup 8 AI agents with Paperclip to help manage his foundation, taxes, newsletter and code (and so can you) > "Setup genuinely took 2 hours, not 2 months $180/month replaces 40+ hours of admin work"