Saving Center
Baseline estimate, actual cost, orchestration cost and estimated saving.
Saving Center
NIZAM is saving-first. Work runs the cheapest way that gets the job done — paid providers are a last resort, only when nothing free will do.
NIZAM first
Deterministic, built-in work — no cost.
Bounded source
Reuse existing packages and references.
Local model
Run a local model on your machine.
Paid — last resort
Only with your explicit approval.
Saving this month
See how much each sprint avoided in paid work.
Provider tiers
Bring your own keys — stored as secure references only.
Budget guardrails
Set a cap; paid work pauses for approval before spend.
How NIZAM works
NIZAM runs work in a strict order, cheapest first. A paid model is a helper for bounded tasks — never the engine, and never the source of truth.
NIZAM deterministic system first
Built-in planning, structure and checks. No model, no cost.
Bounded source (Open Claw) when needed
Reuse allowed, provenance-checked sources and packages.
Local NIZAM Gemma / local provider next
A local model on your machine before anything paid.
Paid providers only as a last resort
OpenAI, Claude, Google, Groq or OpenRouter — only with your explicit approval.
Paid is a helper, not the truth writer
A paid model can draft a bounded sub-task. NIZAM still owns intake, quality checks, scoring and the ledger of truth.
Secure references only
Connect your own provider as a secure reference at a clear boundary. Your credentials are never shown, never logged, and never stored in plain form — only the secure reference is used.
Marketplace & payments
The marketplace and payments are parked for now — nothing is sold or charged yet.
Build for less, ship sooner.
NIZAM is in public preview. Join the waitlist or request early access.