Advanced mathematical optimization for operations

Optimize through mathematics. Scale through precision.

sirat.space uses advanced mathematical optimization to solve complex resource planning problems. Balance multivariable constraints, model network effects, and discover solutions that scale.

Can't wait? Try the playground free for 1 hour — no signup, pre-loaded with sample data.

Multivariable optimizationNetwork mesh architectureReal-time constraint solving
sirat.space/dashboard

8

People

6

Projects

24

Tasks

12

Solved

Capacity — Crew plan

Branches
People
Schedules
Linear ProgrammingNetwork Flow AlgorithmsConstraint SatisfactionGeometric OptimizationMulti-objective SolvingGraph Theory ApplicationReal-time ComputationAdaptive Machine LearningLinear ProgrammingNetwork Flow AlgorithmsConstraint SatisfactionGeometric OptimizationMulti-objective SolvingGraph Theory ApplicationReal-time ComputationAdaptive Machine Learning

How it works

From a spreadsheet guess to a tested, provable plan — in five steps.

No jargon, no black box. Here's exactly what happens when you use sirat.space, screen by screen.

1

Add your people and projects

Set who's available, when, and at what cost. Add projects and tasks with the hours, headcount, and skills each one demands — no spreadsheets, one shared dataset.

sirat.space/people

Amara Okafor

Engineer · 5d/wk

$62/h

Ben Carter

Planner · 5d/wk

$48/h

Chen Wei

Analyst · 4d/wk

$55/h
sirat.space/schedule
Minimize cost · 4 weeksSolve
Week 182%
Week 291%
Week 368%
Week 495%

✓ Optimal — every task fully staffed

2

Solve in seconds

Pick a date range and hit Solve. The optimizer assigns hours per person per day — lowest cost, or highest value, your call — and flags anything it can't fully staff.

3

Branch to test alternatives

Try a reorg, a new hire, a delayed launch — as its own branch, at zero risk to the plan you're already running. Branch off a branch as many times as you need.

Base
Reorg planDelayed launch
+ new hire

Each branch solved independently — nothing else changes until you promote.

Current plan

$18,400

312 hours

Reorg + hire

$16,900

289 hours

Promote “Reorg + hire” →
4

Compare and promote the winner

See exactly what moved — hours, cost, per-person differences — then promote the branch that wins. The rest simply never happened.

5

Or just tell A.C.E. what you need

A.C.E. is a built-in AI assistant that turns a plain-English request — “we're hiring 1 to 20 engineers, model each headcount” — into a full branch tree: people, projects, tasks, and a solve, all built for you. It always works in its own disposable branch underneath the one you're viewing, so a bad prompt never touches your real data.

A.C.E. — { scenarioName }

$ hire 1 to 20 engineers, branch each headcount

Created “A.C.E (3f2a9c1d)” under “{ scenarioName }”…

Built a 20-branch chain, +1 engineer per branch…

Started a batch solve across all 20 branches…

Model layers

One platform. Pick the layer that matches your decision.

A model is a layer you subscribe to. Selecting one swaps the data you work with — everything else, from branching to solving to reporting, stays exactly the same.

Every layer shares the same workspace, branching and solver — only the data it takes changes.

The decision

Who works on what, when — at the lowest cost that still meets demand.

Its data layer

  • Locations

    Shifts, working days, holidays

  • People

    Skills, availability, cost per hour

  • Projects & Tasks

    Demand hours, skills, priority

What you get back

  • Assignments

    Person → task, hour by hour

  • Utilisation

    Per person, against capacity

  • Cost

    Total and per project

Branch itDemand +10% across every projectSolve and compare

SheetSpace

Spreadsheets with branches, diffs and an audit trail.

Our second product: the numbers your business runs on, in one governed place. Branch a book to try a change, review the diff, merge what works — with cell-level history and role-based access down to the branch.

  • Branch and merge spreadsheets like code
  • Cell-level history, diffs and rollback
  • Validation rules that stop bad data at entry
  • Role-based sharing per book and branch

SheetSpace Connect

Your sheets, solved.

Connect a dedicated SheetSpace book to a sirat.space model. Its branches mirror your scenarios, so you model in sheets and get an optimised answer back.

  1. 1MapPair your sheets with the model's data layer.
  2. 2Push & solveSend the data and run the optimiser from your book.
  3. 3PullResults land back as read-only sheets on the same branch.

One integration, every model layer — licensed per installation.

Licensing

Plans for every scale of optimization

Priced per user. Every plan includes the same optimisation engine, with capacity and governance increasing as you grow.

Included with every plan: a model integration and SheetSpace, free for your first 3 months.

Yearly saves 25%

Scientist

For students and educators exploring optimization at no cost.

Free

Per user, forever

  • A model integration + SheetSpace free for 3 months
  • 60 people, projects, tasks & locations combined per branch
  • 30 branches
  • 20 schedules per branch
  • 100 solves per day
  • 7-day solve retention
  • 1-month plan ahead
  • 2 observation log channels
  • A.C.E. AI assistant — 10 AI credits/day free
  • Community Hub access
Choose Scientist

Explorer

A single-user license for one planner who needs branch-based planning with daily solving.

£18.75/user/mo

Billed annually · £225.00 per user/yr

  • A model integration + SheetSpace free for 3 months
  • 1 user
  • 200 people, projects, tasks & locations combined per branch
  • 100 branches
  • 50 schedules per branch
  • 120 solves per day
  • 14-day solve retention
  • 6-month plan ahead
  • 5 observation log channels
  • A.C.E. AI assistant — 25 AI credits/day free
  • Community Hub access
Choose Explorer

Crew

Popular

For collaborative planning teams that require shared access and enterprise readiness.

£13.50/user/mo

Billed annually · £162.00 per user/yr

10 users included · £135.00/mo for the team

  • A model integration + SheetSpace free for 3 months
  • 10 users
  • 800 people, projects, tasks & locations combined per branch
  • 1000 branches
  • 50 schedules per branch
  • 1,500 solves per day
  • 30-day solve retention
  • 6-month plan ahead
  • 15 observation log channels
  • 2-hour playground for team training
  • A.C.E. AI assistant — 60 AI credits/day free
  • API access
  • Team collaboration
  • SLA and onboarding
  • Community Hub access
Choose Crew

Satellite

The decision layer: API-first integration for systems that need direct access to the optimizer.

£243.75/user/mo

Billed annually · £2925.00 per user/yr

  • A model integration + SheetSpace free for 3 months
  • 1 user
  • 1,200 people, projects, tasks & locations combined per branch
  • 1000 branches
  • 50 schedules per branch
  • 2,000 solves per day
  • 14-day solve retention
  • 6-month plan ahead
  • 5 observation log channels
  • A.C.E. AI assistant — 150 AI credits/day free
  • API access
  • SLA and onboarding
  • Community Hub access
Choose Satellite

Custom Order

For bespoke deployments, advanced governance, and tailored operating requirements.

Contact support

  • A model integration + SheetSpace free for 3 months
  • A.C.E. AI assistant — unlimited free AI credits
  • API access
  • Team collaboration
  • Community Hub access
Choose Custom Order

About

Built for decisions that shouldn't be a coin flip.

Mission

Most operational decisions are made once, on incomplete information, and lived with regardless. We think every reorg, hire, or delayed launch deserves to be tested against its alternatives mathematically before it becomes irreversible — not guessed at in a spreadsheet.

The tool

sirat.space models people, projects and tasks as a constraint system, then solves it for the lowest-cost or highest-value schedule. Branch into alternate timelines to try changes risk-free, compare outcomes side by side, and promote the one that wins — all from one dataset, no exports, no spreadsheets to reconcile.

The technology

A mixed-integer linear solver (HiGHS) finds the true optimum, not a heuristic guess. Every branch is a copy-on-write branch — storing only what changed — so forking a timeline is instant and scales cleanly no matter how large the underlying dataset. Everything runs on UTC, with a public API for programmatic access at scale.

Optimization engineered for teams that demand mathematical precision and operational scale.