CognomineCognomine
How it works

Block model and constraints in, schedule out

Cognomine takes a block model and the constraints that shape the plan, and returns a scheduled long-term plan in minutes. Here is what happens between upload and export.

The steps
  1. 1

    Upload

    Bring a block model in a standard format, up to 500,000 blocks. Add the constraints that shape the plan: sequencing rules, capacity limits, grade and blending targets, capital and operating assumptions, and the commodity price decks you want to test against.

  2. 2

    Solve

    Cognomine encodes the model and constraints and solves in the cloud. A run returns in minutes rather than the hours or days a long-horizon schedule usually takes, and it runs on our infrastructure, so there is no compute for you to buy or manage and no extra cost per run.

  3. 3

    Review and compare

    Read the schedule and its supporting outputs. A run costs nothing beyond the credit on your account, so you can change an assumption, solve again, and compare scenarios side by side instead of committing to the first plan that solves.

  4. 4

    Export

    Take the schedule and its outputs into your existing workflow. Cognomine produces the plan; your tools and your judgement decide what to do with it.

What we mean by more detail

The constraints a long-horizon solver usually cannot hold

A long-horizon model is normally trimmed back to stay solvable in reasonable time. That trimming is a compute constraint, not a planning one. Cognomine is built to hold more of it in a single solve, including:

Holding more of these in one solve is what produces a schedule that arrives closer to publishable, so your review time goes into judgement rather than reconciliation.

The engine

Quantum-ready, running today

Cognomine is built on a quantum-ready compute engine. It encodes a planning problem the way a quantum system would, then solves it on today's GPUs. The speed is available now, on standard hardware, with a path to quantum processors as they mature and no change to how the problem is encoded. You do not need to know any of this to use Cognomine. It matters only because it is why the solve is fast.

Benchmarks

What we have measured so far

We have benchmarked Cognomine on realistic strategic scenarios. Across those runs it returned practical schedules in 10 to 20 minutes. On the same problems, open-source solvers had not converged when we stopped them at 12 and 24 hours.

Two things worth saying plainly. We benchmark against open-source solvers, so these figures show the gap against freely available tools, not against every commercial package. And the value is not only speed: because the model holds more of the real constraints, the schedule is closer to something operations can act on with less rework.

Deployment and data

Cloud, kept in Australia, yours alone

Cognomine runs in the cloud, so there is nothing to install and nothing to maintain. Your data is handled on three commitments:

Note: some enterprises may need to run planning tools entirely inside their own environment, offline. If that is you, talk to us about what an on-site deployment would require.