Intelligence, built for decisions.

Connect optimization models, AI agents, and reusable software to the decisions your operation needs to make.

What we mean by Operational Intelligence.

The ability to decide, act, and learn from the world. Each cycle gives the next decision more context.

  1. Decide

    Compute a plan within real objectives and constraints.

  2. Operate

    Put the plan to work, with people in control.

  3. Learn

    Retain outcomes, overrides, and the context behind them.

  4. Decide better

    Bring that experience into the next decision.

Optimization Foundation Model

Generate candidate solutions, refine them through search, and adapt the model with selected experience.

Example
A routing problem with 19 nodes, five robots on a 10 by 10 warehouse grid with obstacles, or chiplet placement and wire routing on a 13 by 13 interposer can be chosen. A learned representation and generator create three candidate solutions. Improvement search revises each through six illustrated states. Selected experience is retained in a buffer and informs a separate, controlled fine-tuning process.
Learned generationImprovement searchSelected experienceRouting example · 3 candidates

Generating candidate solutions

01 / Generate

A learned representation encodes the problem. The generative model proposes three different starting solutions for the same set of nodes.

  1. Generate: represent the industrial problem and generate several candidate solutions.
  2. Search: refine each candidate through a sequence of changes against the objective and constraints.
  3. Fine-tune: retain selected experience, then update representation, generation, and search in a separate, controlled training stage.

Illustrative examples from Omelet’s technical overview, generated once from fixed data rather than a live solver run. Model support and evaluation criteria are defined for each problem.

One intelligence core. Many industrial applications.

One optimization foundation model supplies four solver families. Each decision capability is described by an OCP and packaged as an OCP Server, so applications in different industries run on the same intelligence and one shared operational platform, OIP.

One foundation model. Four solver families: routing, scheduling, packaging and inventory control. Eight OCP Servers and illustrative applications across industries on one common operational platform, OIP.

  1. Foundation Model

    One model, four problem classes. Routing, scheduling, packaging and inventory control each expose it as a Solver API.

  2. Extending with OCP Servers

    An OCP describes a capability. An OCP Server packages it, so the same routing capability serves a fleet or a robot without a new solver.

  3. How an OCP Server works

    Data and rules in. The server formulates the problem, calls its Solver API, and returns a plan for people to review. OIP runs and audits the servers.

  4. OI Factory and OI-Studio

    OI Factory builds and validates capabilities inside Omelet. OI-Studio is where teams explore them: set objectives, run scenarios, compare plans.

Illustrative capability map, not live products or deployments. For the servers available today, see Products.

Agents + optimization

Connect your intent to an optimization capability. Review each proposed plan against your objectives and constraints.

A person asks in natural language. Retrieval and formulation prepare the request within OIP. An OCP Server calls its solver. Analysis presents the result for human review, and reviewed preferences inform a later request.

Three assets. Three systems.

Solve an operational problem, capture reusable intelligence, and compose it into the next application.

OI Factory builds reusable assets. OI-Studio explores them. OI Foundry composes customer-specific applications on OIP.

OI Factory builds reusable capabilities from operational problems. OI-Studio explores scenarios with them. OI Foundry assembles the customer’s application.

Explore the assets and systems

Assets

OCP
The optimization capability, packaged as an OCP Server.
Application Kit
Reusable interfaces and workflow components.
OIP
The shared foundation for operating applications and services.

Systems

OI Factory
Define, solve, validate, and generalize real problems.
OI-Studio
Explore capabilities and compare operational scenarios.
OI Foundry
Compose validated assets around a customer’s operation.

Illustrative workflow. Studio exploration and Foundry delivery can operate independently.

Bring intelligence to your next decision.

Explore the research Explore applications How we work