Pay for what the task needs
Simple work runs on fast, low cost models. Hard work goes to stronger ones.
Auto routing
The right model for every task, automatically.
Each task goes to the lowest cost model that meets its quality bar. If that model is slow or down, work moves to the next one.
Task Summarize vendor contract
- Fast model Lowest costSlow right now
- Balanced model Meets the quality barRouted here
- Advanced model Higher costNot needed
Fallback is automatic. The run keeps going and the switch is logged.
Lower cost without lower quality.
Runs keep going
When a provider has a problem, the task falls back to another approved model instead of failing.
Every choice is visible
The run trace shows which model handled each step and why it was picked.
Lowest cost that meets the bar
Every task carries a quality bar. Auto routing compares the approved models and picks the lowest cost one that can meet it.
Task Summarize vendor contract
- Fast model Lowest costSlow right now
- Balanced model Meets the quality barRouted here
- Advanced model Higher costNot needed
Fallback is automatic. The run keeps going and the switch is logged.
Automatic fallback
If the first choice is slow or unavailable, the task moves to the next approved model and the run carries on. The switch is recorded in the trace.
- No failed runs because one provider is down
- Fallback only to approved models
- Every switch is logged
Workspace
Run trace
Request received Complete Policy matched Complete Usage summary checked Verified Legal clause compared Verified Approval requested Waiting
Inside your model policy
Routing only ever chooses from the models your policy allows for that team and agent, so cost savings never bypass governance.
Low risk draft Fast model
Legal reasoning Approved advanced model
Sensitive data Private provider
Central policy
Controls cost, quality, latency, privacy, and task complexity.
How it works
From request to recorded result.
-
A task starts
It carries the quality bar for the work.
-
Models are compared
Only the approved ones for that team and agent.
-
The best fit is picked
The lowest cost model that meets the bar.
-
Fallback if needed
Slow or down? The next model takes over.
-
The choice is logged
The trace shows which model ran each step.
Every feature runs inside the same six controls.
Identity, access, approvals, supervision, audit and policy apply to every agent and every run, whichever part of the workspace it happens in.
See governance and audit- Identity
- Access
- Approvals
- Supervision
- Audit
- Policy
FAQ
Auto routing: questions and answers
Short answers to what teams ask before they put this part of SoftworkerAI to work.
Contact us directlyHow does auto routing pick a model?
It looks at the approved models for that team and agent and picks the lowest cost one that meets the quality bar for the task.
What happens if a model is down?
The task falls back to the next approved model and the run continues. The switch is recorded in the run trace.
Can routing choose a model we have not approved?
No. Routing only chooses from the models your policy allows.
Will routing lower the quality of the work?
No. Routing only picks models that meet the quality bar for the task. It saves cost on simple work, not on hard work.
The multiplayer workspace for humans and AI agents is how teams move beyond prompting and safely hand AI real work. It gives people one place to plan, approve, and oversee execution, so they can run work with AI agents they can govern and trust.