Intelligence

Decisions you can inspect.

The optimisation and scheduling layer that explains why one eligible region or provider should win over another.

System view

This is deterministic optimisation against demonstration data. It is not presented as autonomous or opaque AI.

Decision model

Change the objective. Read the result.

Each objective changes factor weights, eligibility and the final ranking. Rejected regions remain visible so the outcome is understandable.

Policy demo. Candidate regions, scores and conclusions use illustrative data and do not represent a current FabriX production decision.

Objective weightsPolicy demo

No single axis dominates. The scheduler trades cost against latency and carbon until the marginal gain on each is equal.

Candidate regions8 of 8 example regions eligible

Paris clears every constraint at once: 6 ms, $2.19/hr and 56 g/kWh with headroom to grow.

CostLatencyCarbonAvailabilityResidency

Placement factors

Some inputs are hard constraints. Others contribute to a score. The policy should make that distinction visible.

Compare the modelled price of eligible allocations, including the effect of region and contract term.

Prefer regions that keep the request inside its network latency limit.

Reject supply that does not match the required architecture, memory profile or interconnect.

Score capacity against the requested quantity and the operational state of each region.

Use electricity price, renewable share and carbon intensity when the workload can move in time or geography.

Filter out locations that do not satisfy residency, jurisdiction or compliance policy.

Prefer allocations that can tolerate endpoint or provider failure when continuity matters more than price.

Tell FabriX which technical, geographic and commercial inputs should govern your compute requests.