OpEra is live. Custom AI decision engines for operations teams: production schedules, warehouse plans, delivery routes, re-planned when a machine breaks, a truck runs late or an order changes. 38% lower delivery cost for a major retailer in weeks. Build at MIT. YC F26.

Danish ferry line Molslinjen forecasts every departure a year ahead, then sets ticket quotas, fares and how to pack the car deck. Estimated gain: up to $3.2M a year, 3% less fuel. Setting capacity and prices close to the date? OpEra builds that model in weeks, on your own data.

Chevron runs seven refineries on one optimization model: which crude to buy, what to make, how to run each unit. Chevron puts its value at close to $1B a year. Every plant makes those three choices. OpEra builds that model in weeks, on the data you already have.

Meituan delivers 60 million orders a day. Every 30 seconds a model decides which courier takes which order. Delivery time per order fell 21%, courier distance 24%. A dispatch desk with 40 drivers has the same problem. OpEra builds that model in weeks, on your own data.

Air New Zealand's crew rosters took six people two weeks to build. Optimization models now do the maths in under an hour, saving about NZ$15.7M a year by the authors' estimate. Still building rosters by hand? OpEra builds that model in weeks, on your own data.

Intel plans which chips to offer and where to make them as one problem, across 200+ factories. Its count: $1.9B more revenue and $1.5B lower costs a year since 2009. A plant with a few hundred products has the same problem. OpEra builds that model in weeks, on your own data.

Walmart built a model that plans each truck's route and how its trailer is loaded together. In fiscal 2023 it cut 108,000 routes and saved $91.5M, by Walmart's count. A 30-truck fleet has the same problem. OpEra builds that model in weeks, on your own data.

Microsoft planned its cloud hardware in spreadsheets. A model now picks the supplier, data center and date for each request, and cycle times were cut in half. Is your plan still a spreadsheet reworked by hand? OpEra builds that model in weeks, on the data you already have.

A snowstorm cancelled 112 Continental flights. A model re-planned the crews in three and a half minutes. By hand it took hours. When a machine or a driver drops out of your day, you need a new plan in minutes too. OpEra builds that re-planning model in weeks, on your own data.

See the decision change · Season 4, Episode 1: The monthly allocation plan How much of what, from where, to whom? Prices move mid-month, a big customer changes its order, and the month is re-planned in seconds. An illustration of what OpEra builds on your data, in weeks.

Before changing a driver policy, Schneider National tested it in a model. The plan would have cost $30M a year. The model found most of the benefit for $6M. Your fleet can test a decision before paying for it. OpEra builds that model in weeks, on your dispatch desk's own data.

See the decision change · Season 3, Episode 1: The dispatch board Which technician goes to which job, and when? Three emergency calls land at 10 a.m., and the whole day is reassigned in seconds. An illustration of what OpEra builds on your own data, in weeks.

Taco Bell stopped staffing shifts on a budget and a gut feel. A model set who starts when, from 15-minute forecasts, and saved about an hour of labor per restaurant per day. A warehouse or a plant has the same problem. OpEra builds that staffing model in weeks, on your data.

See the decision change · Season 2, Episode 2: The changeover sequence Which product runs on which line, in what order? The biggest customer doubles an order due Friday, and the week is resequenced in seconds. An illustration of what OpEra builds on your own data, in weeks.

Waste Management drew its routes with highlighters and string. Then a model re-sequenced every stop and took 984 routes off the road in nine months. You don't need 26,000 trucks for that. OpEra builds that routing model for your fleet in weeks, on your own data.

Procter & Gamble asked one question across its North American plants: which plant should make what? The answer took more than $200M a year out of its costs. The same question exists with three plants. OpEra builds the model that answers it, in weeks, on your own data.

Google let a model choose the settings of a data centre's cooling plant and used up to 40% less energy for cooling. Operators stayed in charge. Every plant has settings someone adjusts by experience. OpEra builds the model that picks them, in weeks, on your own data.

See the decision change · Season 2, Episode 1: The production schedule Which job runs on which machine, and when? The paint line goes down for six hours, and the whole week is rescheduled in seconds. An illustration of what OpEra builds on your own data, in weeks.

Coca-Cola Enterprises put an optimization model under its 300+ dispatchers and about 10,000 trucks. Delivery cost fell by $45M a year, and the dispatchers stayed. It works for 20 trucks too. OpEra builds that routing model in weeks, on the data your dispatchers already have.

See the decision change · Season 1, Episode 2: The delivery routes Which truck takes which stops, in what order? A driver calls in sick at 6 a.m., 40 stops have no truck, and every route is redrawn in seconds. An illustration of what OpEra builds on your own data, in weeks.