Beyond Optimization Engines: Why Harumi Represents a New Generation of Decision Platforms

Optimization has long been essential for industries such as logistics, manufacturing, workforce management, and planning. Traditionally, these problems were solved by powerful engines built around advanced algorithms, heuristics, and constraint solvers.

Timefold is a strong representative of this generation. Harumi, however, reflects a shift in how optimization is designed, used, and adopted across organizations.

Both platforms solve optimization problems but they differ fundamentally in how they approach AI, usability, deployment, and extensibility

Timefold: An Optimization Engine Built for Flexibility

Timefold is built on great research in planning and scheduling optimization. Its core strength lies in its solver technology, which efficiently explores large solution spaces using deterministic and heuristic algorithms, often referred to as Planning AI. Key characteristics of Timefold include:

• High-performance constraint solving and scheduling

• Strong flexibility through API-based integration

• Freedom to design and build fully custom user interfaces

• Excellent fit for deeply embedded optimization use cases
  
  

Timefold provides, for example, pre-made planning and scheduling models for common problems. These models are robust and production-ready, allowing teams to start quickly when their use case aligns closely with existing templates.

Organizations can design any interface they want, select or adapt a pre-built model, and simply “plug in” the optimization engine. However, this flexibility assumes:

• Dedicated engineering resources

• Optimization expertise

• Longer iteration cycles tied to development workflows

In practice, Timefold works best when optimization is treated as a backend capability managed by technical teams.

The Challenge: Optimization That Doesn’t Scale Organizationally

While Timefold excels at solving hard problems, many companies face a broader challenge:

Optimization power does not automatically translate into adoption.

When optimization depends on:

• Custom interfaces

• Technical configuration

• Specialized knowledge

it often remains limited to a small group of experts. This is where Harumi takes a different path.

Harumi: An AI-Native Optimization Platform

Harumi was designed with a different premise: optimization should be usable by both technical and business teams.

Rather than positioning optimization as an engine to be embedded, Harumi presents it as a platform and execution environment — combining mathematical optimization with AI-guided interaction.

Key Differences That Matter in Practice

1. AI That Assists People, Not Just Decisions

Timefold’s “AI” is algorithmic — rooted in planning heuristics rather than generative AI. Harumi complements optimization with conversational AI, allowing users to:

• Describe goals and constraints in natural language

• Get guidance while building models

• Interact with optimization results intuitively

This shifts optimization from a purely technical activity to a collaborative and strategic process.

2. Knowledge Barrier and Accessibility

Timefold assumes a solid understanding of optimization concepts and system integration. Harumi deliberately lowers the knowledge barrier:

• No formal OR background required

• Low-code modeling workflows

• AI guidance throughout the lifecycle

This enables operations and business teams to engage directly with decision models instead of relying entirely on technical intermediaries.

3. Pre-Made Models and Extensibility

Both platforms offer pre-made optimization models, but they treat them differently.

• Timefold provides pre-built models that teams can integrate and extend primarily through code and application-level customization.

• Harumi also offers pre-built models, but has the feature to create new ones personalized for each business.

In Harumi, users can:

• Clone existing models

• Use them as "skeletons" to create new optimization solutions

• Extend and adapt models directly within the platform
  
  

4. Customization: Same Power, Different Experience

Both platforms offer very high customization, but in different ways:

• Timefold: Customization happens primarily through building custom applications and interfaces around the engine.

• Harumi: Customization happens directly within the platform — users customize models, constraints, and workflows without building a separate application.

Harumi preserves flexibility while removing unnecessary complexity.

5. Deployment and Operations

Timefold is typically deployed as an optimization service integrated into existing systems. Harumi works as an execution environment:

• No deployment steps

• No infrastructure management

• Models run directly in the platform

This reduces operational overhead and shortens time-to-value.

6. Typical Usage Patterns

• Timefold: Best suited for organizations with strong engineering teams embedding optimization deep into products or systems.

• Harumi: Ideal for organizations that want fast rollout, collaboration across teams, and optimization embedded into daily operations.
  
  

Conclusion: Optimization Needs to Be Usable to Be Valuable

Timefold remains a fair choice for teams that need a highly flexible optimization engine and are prepared to build everything around it.

Harumi reflects the next evolution of optimization platforms — one that recognizes that AI, interface design, extensibility, and usability are just as important as algorithms.

By lowering the knowledge barrier, removing deployment friction, and allowing teams to evolve models using pre-built skeletons, Harumi turns optimization from a specialized discipline into a practical, everyday decision tool.

In a world where better decisions must be made faster and by more people, Harumi’s AI-native approach offers a compelling vision for the future of optimization.