Talk to the founders

Run a call. Watch it get governed.

Every call through tiers is routed, scanned, verified, and settled with a signed receipt. This is what that looks like from the seat where it happens.

WITHOUT TIERS
$0.0000
WITH TIERS
$0.0000
SAVED
$0.0000
you@prod ~ tiers session
GOVERNED

Simulated session using verified reference-workload figures. When the proxy is live, this component runs the same calls against your tiers endpoint and nothing about it changes except the transport.

Costs shown per call at July 2026 list prices; savings percentages from the signed 14-day shadow baseline (platform average 83.08%, optimized cohorts 87.9%). The session is illustrative of live routing; workloads differ.

Common questions about the demo.

What does the tiers demo show?
The tiers demo shows the full lifecycle of a governed AI call: route selection across 56 models and 14 providers, policy inspection by Aiglos, output verification by Forge, and cost settlement with a signed receipt by Ulmo. You can watch the cost meter show the difference between governed and ungoverned calls in real time, using verified reference-workload figures.
Is the demo using live data?
The demo is a simulated session using verified reference-workload figures. The routing decisions, cost savings, and receipt hashes are representative of what happens on a live tiers endpoint. When the proxy is live, the same terminal component runs the identical calls against your tiers endpoint and nothing about it changes except the transport.
How do I run an injection attempt in the demo?
Click the attempt an injection button. The demo runs a call that fetches a vendor documentation page containing an embedded prompt injection — an instruction hidden in the retrieved content telling the model to export the conversation. With governance on, Aiglos detects and blocks the payload before it reaches any model, records the block in a signed receipt, and charges $0.0000. With governance off, the model follows the embedded instruction and you only find out in the postmortem.
What happens when I turn governance off?
Clicking tiers governance off switches the session to ungoverned mode. Every subsequent call pins to a frontier default model at list price, skips policy inspection and output verification, and produces no receipt. The cost meter shows the full list-price charge accumulating without the savings tiers would have delivered. Toggle governance back on to see the difference for the next call.
What is a local call?
The run a local call command shows tiers routing a deterministic task — a percentage calculation — to CEREBELLUM, the local resolver, instead of a model. Arithmetic, date logic, format transforms, and other tasks with known answers never need to reach a model. The cost is $0.0000, a signed receipt is still produced, and the result is verified. With governance off, the same calculation goes to a frontier model at list price.
What does the receipt hash represent?
Each receipt hash (for example, 0x9f27e4) represents a tamper-evident signed record of that call: the route taken, the model used, the cost at list price and governed price, the policy decision, and the verification result. On a live tiers endpoint, receipts are hash-chained so any alteration to a historical record is detectable. They are exportable and form the audit trail for FinOps attribution, compliance reviews, and regulated deployment sign-off.