Skip to content
Advancing Bits to Master Atoms

Simulation before action.
Plays the options forward before compute or lab time is spent.

The options that cannot reach the goal are set aside early, and the budget goes to the runs most likely to move the work forward.

Plan and decide · one of the fourteen capabilities of MS Duplex

A graphite block carved with a branching tree of channels, a silver sphere at the fork and small cobalt-blue spheres along only one branch.A graphite block carved with a branching tree of channels, a silver sphere at the fork and small cobalt-blue spheres along only one branch.

Your team gets

  • Fewer expensive runs that were never going to work
  • Budget spent where it counts
  • The reasoning behind every choice, on record

What simulation before action means for your team.

Simulation before action means your AI thinks a costly step through before it takes it. When there are several ways forward, MS Duplex weighs the likely outcome of each one against the goal, so the choice rests on evidence rather than on the first idea that came up.

The weighing draws on your team’s own experience: the results that were confirmed, the conditions they hold under, the corrections your experts made and the runs that failed. An option that already failed under the same conditions, or that cannot reach the goal, is set aside before anything is spent on it.

Your own simulators, instruments and analysis code stay as they are, and they still produce the results. MS Duplex helps decide which runs are worth making, and records the reasoning behind each choice.

Why research needs to look before it runs.

Compute, instrument time and expert attention are the real budget of a research team. Every run that was never going to work is spent from it.

Many research decisions come down to choosing the next run from several candidates. A long simulation on shared compute, an experiment that uses scarce material or instrument time, a design review that takes an expert away from other work: each one is worth running only if it can move the work toward the goal.

Without a look ahead, agents choose by habit or by the first plausible idea. Options your team already knows will fail are proposed again, and the budget runs out before the most promising options are reached. Simulation before action puts your team’s confirmed experience in front of every costly choice.

  • Runs are expensive

    A single run can take a large share of the compute, instrument time or expert attention a team has.

  • Budgets are fixed

    Every run spent on an option that could not reach the goal is one fewer run for the options that could.

  • Experience should count

    What your team has already confirmed and ruled out should shape the next choice before it is made.

  • Choices need reasons

    A leader approving a costly run should be able to see why it was chosen over the alternatives.

A satin-silver balance on a graphite base, two graphite blocks in one pan and a single cobalt-blue block in the other.A satin-silver balance on a graphite base, two graphite blocks in one pan and a single cobalt-blue block in the other.

How simulation before action works in your team’s day.

Your agent keeps proposing the next step. Before a costly one runs, MS Duplex looks ahead.

  1. 1

    Options are on the table

    Your agent reaches a costly step and has several ways forward: different settings, candidates, methods or runs.

  2. 2

    Each option is played forward

    MS Duplex weighs the likely outcome of each option against the goal, using the results, conditions and corrections your team has confirmed.

  3. 3

    Dead ends are set aside

    Options that cannot reach the goal, or that already failed under the same conditions, are set aside before anything is spent on them.

  4. 4

    The budget goes where it counts

    The most promising options go ahead through your own tools, and the reasoning behind the choice goes on record.

An example, in any field where runs are costly.

A team’s agent has budget for a small number of expensive runs and a long list of candidate settings to choose from.

  1. 1

    The list

    The agent proposes twelve candidate settings for the next round of runs. The budget allows four.

  2. 2

    The look ahead

    MS Duplex weighs each setting against the goal. Three sit beyond an operating limit an expert confirmed last month, and two repeat a combination that already failed under the same conditions.

  3. 3

    The choice

    Those five are set aside, each with its reason. Of the seven that remain, the four most likely to move the work forward go ahead, and the other three wait for the next round.

  4. 4

    The record

    The results come back and are confirmed. The team lead sees why each setting was run or set aside, and the next round starts from what this one confirmed.

Simulation before action works with the other capabilities.

Looking ahead is only as good as the experience behind it, and every choice it makes adds to that experience.

  • Long-horizon planning

    At each checkpoint of a long campaign, the next options are played forward before the plan commits budget to them.

  • Learning cause and effect

    What your team has learned about which action led to which outcome, under which conditions, is what each option is weighed against.

  • Decisions that know their limits

    Where the evidence clearly favors an option the choice is settled directly, and where it is thin it goes to your model or a researcher.

  • Rules and approvals

    In governed environments, the options that go ahead can still wait for a person’s approval before the costly step begins.

What your team gets.

Research budget spent on the runs that can move the work forward, with the reasoning on record.

  • Fewer runs that were never going to work

    Options that cannot reach the goal are set aside before compute or lab time is spent on them.

  • Budget spent where it counts

    The runs that go ahead are the ones most likely to move the work toward the goal.

  • Your experience in every choice

    Confirmed results, known conditions and expert corrections shape each costly choice before it is made.

  • Known failures set aside

    An option that already failed under the same conditions is set aside instead of being proposed again.

  • Reasoning on record

    Every choice keeps the reasons it was made, so anyone can see why one run went ahead and another waited.

  • Your tools, as they are

    Your simulators, instruments and analysis code still do the work, and MS Duplex helps decide which runs are worth it.

Common questions.

Does simulation before action replace our own simulations?

Your simulations and experiments still produce the results. Simulation before action helps decide which of them are worth running, by weighing each option against the goal and against what your team has already confirmed.

What does MS Duplex use to weigh the options?

Your team’s own experience: the results that were confirmed, the conditions they hold under, the corrections your experts made and the runs that failed. Each choice keeps a record of what it rested on.

What happens when the evidence is too thin to choose?

MS Duplex says so. Where your team’s experience does not yet favor one option, the choice goes to your model or is flagged for a researcher, with the reason.

Can a person still approve a costly run?

Yes. In governed environments, the steps you select, such as an expensive simulation, wait for a person’s approval before they begin, and the approval goes on record.

Make the next run benefit from the last.

Bring us the research your AI should get better at. We will plan a pilot on your own workflows, with your experts as the judges.