Hey HN, I'm Evan. I've been building autonomous agent systems for the past year, and Opensoul is the result of productizing what I learned.
*What is it?*
Opensoul is a pre-configured deployment of Paperclip (an agent orchestration platform) built specifically for marketing. It ships with 6 agents organized as a real marketing agency:
Each agent runs autonomously on scheduled heartbeats. They check their work queue, execute tasks, delegate to teammates, and report progress. Just like a real marketing team—but AI.
*Why I built it:*
Most AI marketing tools give you single-purpose assistants: "write me a blog post" or "generate ad copy." But marketing isn't a series of disconnected tasks—it's coordinated strategy across people, channels, and time.
I wanted to manage marketing outcomes, not prompts. Tell the team "launch product X to 10K signups" and let them figure out the campaigns, content, and optimization.
*Tech details:*
- Built on Paperclip (also open-source: github.com/WeAreBini/paperclip)
- Agents can use any LLM backend (Claude, GPT-4, etc.)
- TypeScript/Node.js, embedded PostgreSQL (zero config to start)
- MIT licensed
*Current limitations:*
- No native social media posting (agents can draft but you approve/post)
- No built-in analytics integrations yet (on the roadmap)
- Requires comfort with CLI/terminal to run locally
Hey HN, I'm Evan. I've been building autonomous agent systems for the past year, and Opensoul is the result of productizing what I learned.
*What is it?*
Opensoul is a pre-configured deployment of Paperclip (an agent orchestration platform) built specifically for marketing. It ships with 6 agents organized as a real marketing agency:
- Director (strategy, team coordination) - Strategist (market research, positioning) - Creative (copy, messaging, brand voice) - Producer (content production, editorial calendar) - Growth Marketer (SEO, acquisition, conversion) - Analyst (measurement, attribution, ROI)
Each agent runs autonomously on scheduled heartbeats. They check their work queue, execute tasks, delegate to teammates, and report progress. Just like a real marketing team—but AI.
*Why I built it:*
Most AI marketing tools give you single-purpose assistants: "write me a blog post" or "generate ad copy." But marketing isn't a series of disconnected tasks—it's coordinated strategy across people, channels, and time.
I wanted to manage marketing outcomes, not prompts. Tell the team "launch product X to 10K signups" and let them figure out the campaigns, content, and optimization.
*Tech details:*
- Built on Paperclip (also open-source: github.com/WeAreBini/paperclip) - Agents can use any LLM backend (Claude, GPT-4, etc.) - TypeScript/Node.js, embedded PostgreSQL (zero config to start) - MIT licensed
*Current limitations:*
- No native social media posting (agents can draft but you approve/post) - No built-in analytics integrations yet (on the roadmap) - Requires comfort with CLI/terminal to run locally
*Try it:*
```bash git clone https://github.com/iamevandrake/opensoul.git cd opensoul && pnpm install && pnpm dev ```
I'd love feedback on the agent architecture. Is this how you'd structure a marketing team? What capabilities would make this useful for your projects?