Sometimes the best product decisions come from the questions you’re willing to ask out loud.
When we first built Anjin, the focus was clear: let people move fast, create agents, and explore freely. We didn’t want the experience to feel gated or restrictive.
But as usage scaled, two things became clear:
- True unlimited access isn’t sustainable - for the user experience or the technical infrastructure.
- Smart limits actually help people use the platform better, not worse.
We’re still working out the best way forward. Here’s where our thinking is today.
What We're Thinking About
Right now, we’re actively exploring a few different models:
- Restricted seats: Limiting how many team members can access an account based on the plan.
- Restricted domains: Tying agents more closely to verified domains to focus agent utility.
- Credit systems: Letting users “spend” on what they value most - more agents, more domains, more swaps.
Each has its strengths - and each raises new questions.
- Seats encourage focus, but might slow down flexible teams.
- Domain restrictions improve clarity, but might constrain early experimentation.
- Credits give flexibility, but add an extra mental step.
No decision is final yet. We're intentionally taking our time.
Why This Isn't Just About Business Models
There’s another big factor driving this conversation: technical load.
Every time an agent runs, it’s not just a database read or a simple script.
Our agents have been built from years of combined technical experience to execute specific, valuable tasks in the real world.
Behind each execution, there’s:
- High-compute processing
- Multi-API coordination
- Secure token management
- Real-time or near-real-time data operations
- Auditable, server-side logging for traceability
Each agent run puts real weight on the backend infrastructure.
It’s invisible to the user (by design) - but from an engineering perspective, it’s heavy lifting.
That’s why being thoughtful about usage isn’t just about pricing - it’s about performance, stability, and ensuring every user continues to get a fast, reliable experience as we scale.
Educating Users as We Build
One thing we do know: clear limits can create more value, not less.
When users understand:
- What actions cost effort (both theirs and the system’s)
- Where they have freedom to explore
- How to build efficiently within the system’s design
...they end up getting more from the platform, not less.
That’s why any model we choose - whether seats, domains, credits, or a hybrid - will come with education baked in.
Because the best users aren’t the ones who exploit the system. They’re the ones who understand it deeply.
What's Next
- We’re gathering real-world user feedback (join us and tell us what matters most to you).
- We’re testing different models internally to see what feels most natural to real usage patterns.
- We’re keeping the system flexible enough to adapt if (and when) new behaviours emerge.
The goal isn't to gate innovation.
It's to create the conditions where innovation thrives.
Final Thought: Limits Create Meaning
We don’t want limits that feel arbitrary.
We want limits that reflect the real value of what’s happening under the hood—and help users channel that value to do meaningful things.
In a world where AI platforms often hide their complexity behind “magic” buttons, we’d rather be transparent: building powerful agents takes serious work, and using them thoughtfully is part of the craft.
Thanks for being on the journey with us.
Got thoughts on the best model?
Join the community, or drop a comment. We’re listening—and building this openly, with you.
Catch up on the story so far:
- Inside Anjin #01: From Lovable to Liveable
- Inside Anjin #02: Why We’re Listening to BigQuery
- Inside Anjin #03: Dinner with AI
- Inside Anjin #04: Under the Hood of Agent Deployment
