Week of February 7: You're in Control
Week of February 7: You're in Control
The most common thing people say about AI is that they don't fully trust it yet. This week, we shipped the features that address that—directly.
What's New
Human-in-the-Loop Approvals
Your atoms can now pause and check in before they act. When an atom reaches a consequential decision point, it surfaces for your review. You read what it's planning to do, approve or adjust, and it proceeds. No surprises. No silent mistakes.
This is the kind of oversight that makes handing off real work to AI actually feel safe. It's off by default and configurable per project—use it where you need it.
Neural Tuning
Your atoms get smarter over time. Neural Tuning is a new settings page that controls how your AI workforce learns from the work it does in your workspace. Adjust the learning rate, review what patterns it has developed, and guide how it improves.
The goal: atoms that feel less generic and more yours, the longer you use them.
A New Subscription Experience
The path from demo to a full Supanova workspace is now clean and fast. The new billing flow handles the transition in minutes—workspace stays intact, nothing resets, you're just unlocked. Upgrading is no longer a project in itself.
Documents Worth Sending
Deliverables got a full redesign. When your atoms produce a document, it now comes out looking like something you'd actually share with a client or present in a meeting. The output matches the quality of the work inside it.
Risk Calculators
Two new assessment tools are now live in the resource library: the Observability Tax Calculator and the Fleet Risk Calculator. Drop in your numbers and get a clear picture of exposure and cost. They're fast, shareable, and built to spark the conversation that leads to action.
Codex Quick Paste
Getting content into your knowledge base just got easier. Quick Paste lets you drop text directly into Codex without going through the full document flow. Capture the context, move on.
Under the Hood
When a task is deleted, its subtasks no longer disappear with it—they get properly reassigned so nothing falls through the cracks. Real-time comments are now authenticated end-to-end, keeping workspace conversations private and reliable.
Thirty-six weeks in. This week was about trust: giving you more control, more visibility, and a system that earns confidence through transparency. Keep building with us.