4 comments

[ 2.1 ms ] story [ 21.5 ms ] thread
Since it's day 1, a lot of things are still rough.

I'm intentionally building this in the open so people can see the architecture evolve and suggest better approaches.

If anyone wants to contribute or experiment with ajne.dev, I'd love PRs.

One idea I'm exploring is letting people define custom "AI employees".

  Cool concept. The multi-agent approach with role-specific AI employees (CEO, CTO, CMO) is interesting — reminds me of CrewAI but more startup-focused.                                                       
                                                                                                                                                                                                               
  A few thoughts:                                                                                                                                                                                            
                                         
  - How do the agents handle disagreements? For example, if the CTO agent wants to build a monolith but the CEO agent pushes for microservices based on scaling assumptions. Is there a conflict resolution
  mechanism?
  - The "custom AI employees" idea has legs. Different startups need different roles — a SaaS company needs a DevOps agent, an e-commerce one needs a supply chain agent.
  - For day 1 this is solid. Architecture suggestion: consider adding a shared context/memory layer so agents can reference what other agents decided. Without that, each agent works in a silo.

  Good luck with the project.
Since this is open source, I'm still debating what Ajen should ultimately become.

One direction is turning it into a framework for building multi-agent companies, where developers can define roles, coordination logic, shared memory, and decision systems between AI employees.

The other direction is making it a product for founders, where you describe a startup and AI employees (CEO, CTO, CMO, etc.) collaborate to plan and build the company — writing product specs, designing the architecture, shipping the product, and planning the GTM strategy.

I'm curious which direction would be more interesting:

• Infrastructure for building AI organizations • A practical AI co-founder product for founders

Also wondering whether the system should eventually include roles like finance, legal, operations, etc., or if it's better to keep the core roles minimal and let people define custom AI employees.