Blog / October 08, 2024

AI that plans, thinks, and acts: What to know about reasoning engines

Ryan Brelje, Senior Product Marketing Manager

What is a reasoning engine?

The field of AI has long grappled with a fundamental question: Can machines truly reason? Recent advancements in agentic AI are providing compelling answers.

Unlike traditional AI systems that follow predefined rules, agentic AI is designed to think and act autonomously. At the core of this approach is the reasoning engine — a sophisticated AI system that goes beyond data processing to tackle complex challenges through logical analysis.

A reasoning engine enables AI to understand problems, develop strategies, and execute solutions independently. This capability is transforming how businesses approach task automation and problem-solving. In this blog, we'll explore the mechanics of reasoning engines, their role in agentic AI, and how they're changing the landscape of enterprise operations.

Read more about the Moveworks Reasoning Engine.

What is a reasoning engine?

A reasoning engine is an AI system that understands a user's objective, develops a plan to achieve the objective, executes function calls according to its plan, evaluates the success of this execution, iterates on the plan, till it successfully achieves its original objective. In some ways, it attempts to mimic how human beings solve problems.

In practical terms, a reasoning engine allows the system to dynamically figure out how to help a user, rather than follow a set of pre-defined rules or decision trees. It serves as the brain of AI assistants.

How does a reasoning engine work?

The functionality of a reasoning engine revolves around its ability to understand, plan, and act. It processes user inputs, analyzes them within their given context, and determines the most effective solution. Here are the key components that enable this process:

  1. Natural language processing (NLP): It starts by interpreting user queries, understanding not just the words but the intent and context behind them.

  2. Knowledge graph: A structured database that organizes and links information from various sources, allowing the engine to quickly access relevant data.

  3. Inference engine: The logical core that applies reasoning rules to the available data to reach conclusions and make decisions.

  4. Machine learning models: These components continuously improve the engine's performance by learning from new data and interactions.

  5. API Integrations: Connections to various enterprise systems (like ITSM, HRIS, and ERP) that allow the engine to access and manipulate data across the organization.

What makes a reasoning engine so effective?

In a world where AI often stops at responding, a reasoning engine acts. This ability to act autonomously on complex tasks distinguishes it from simpler AI systems. Here's what makes it particularly effective:

  • Contextual analysis: Every user, role, and request is different. A reasoning engine tailors its actions based on these factors, ensuring solutions are specific and relevant.
  • Intelligent source selection: It doesn’t just pull data randomly. The engine can identify and prioritize the most reliable and relevant information sources for each task.
  • Multi-step planning: Complex issues often require more than one action. The engine breaks down problems into clear steps, executing each seamlessly.
  • Cross-system orchestration: It acts across multiple systems at once, bridging gaps and unifying processes in real time.

The Moveworks Reasoning Engine: Agentic AI in action

The Moveworks Reasoning Engine is the cornerstone of our Copilot platform, embodying the principles of agentic AI to revolutionize enterprise problem-solving. 

But what exactly is agentic AI, and how does our Reasoning Engine apply this concept?

Understanding agentic AI

Agentic AI refers to AI systems capable of autonomous action to achieve complex goals on behalf of users. Instead of following predefined logic or conversation flows, these systems can independently reason their way toward specific objectives. This approach allows AI to function more like a human assistant, understanding challenges and devising solutions proactively.

The Moveworks approach to reasoning

Our Reasoning Engine leverages this agentic AI framework to transform how enterprises handle everyday employee challenges. Here's how it works:

  1. Advanced LLM integration: The engine combines multiple large language models (LLMs), each fine-tuned for specific tasks such as reasoning, planning, and summarization. This layered approach enables sophisticated problem-solving capabilities.

  2. Modular problem-solving: The AI breaks down complex issues into smaller, manageable steps. It then executes these steps sequentially using various plugin capabilities, allowing it to operate across different enterprise systems.

  3. Diverse reasoning techniques: Our framework applies different LLMs to various reasoning and decision-making methods. 

  4. Goal-oriented framework: The Copilot identifies user goals and systematically determines the steps needed to achieve them. This comprehensive approach makes it adaptable to a wide range of employee support scenarios.

In practice, the Moveworks Reasoning Engine functions much like a skilled personal assistant. When an employee presents a challenge, our Copilot:

  1. Understands the context and specifics of the issue

  2. Performs necessary research and analysis

  3. Develops a tailored solution

  4. Executes the solution across relevant systems

This process dramatically reduces resolution times for support issues. By automating these tasks, the Reasoning Engine frees human agents to focus on more complex, high-value work, ultimately enhancing operational efficiency across the organization.

The result is a versatile, intelligent system capable of addressing a broad spectrum of employee support needs efficiently and effectively.

How the Moveworks Reasoning Engine works

Building on the agentic AI framework we discussed earlier, let's explore how the Moveworks Reasoning Engine operates in practice. The engine's problem-solving process can be broken down into three main stages: understanding, planning, and executing. This approach allows our Copilot to handle a wide range of employee support scenarios effectively.

Understanding user needs

The process begins the moment a user interacts with the Copilot. The Reasoning Engine:

  1. Analyzes the user's query to identify their specific needs

  2. Interprets the message to grasp the topic and subject matter

  3. Aims to clarify the root issues and potential solution paths

  4. Recalls previous conversations for relevant context

  5. Rewrites the query in a format it can process, applying past learning experiences

This comprehensive understanding stage sets the foundation for effective problem-solving.

Planning for success

Once the engine understands the user's needs, it develops a targeted action plan:

  1. Creates a step-by-step sequence of plugin calls

  2. Uses topic data and the rewritten query to filter its plugin catalog

  3. Selects the most relevant plugins and application-specific actions

  4. Identifies the systems to interact with and the actions to take

The plugin system is crucial here. It allows the Copilot to interact with hundreds of different applications and perform thousands of actions. Our selection model, built on extensive employee support expertise, efficiently pre-selects the most appropriate plugins for each scenario.

Executing with purpose

With the plan in place, the Reasoning Engine moves to action:

  1. Systematically activates the selected plugins

  2. Performs specific tasks such as searching knowledge bases, analyzing related files, and accessing relevant support forms

  3. Executes these actions rapidly, often in fractions of a second

  4. Generates a concise, cited summary of the most relevant findings

  5. Adapts the plan if necessary, either by using different plugins or requesting more information from the user

This flexible execution mimics human problem-solving, allowing the engine to handle complex and varied scenarios effectively.

Advanced features of the Moveworks Reasoning Engine

Agentic architecture

The engine's agentic architecture coordinates multiple LLM agents to tackle complex tasks from simple prompts. This design:

  • Transforms natural language inputs into powerful, generative workflows
  • Enables the Copilot to adapt dynamically to various support scenarios
  • Allows for seamless execution across multiple enterprise systems

Frontier large language models

Our Reasoning Engine leverages a diverse ecosystem of LLMs:

  • Includes global, proprietary, and customer-specific models
  • Allows for comprehensive addressing of employee issues
  • Features a flexible design that can integrate emerging models
  • Ensures continuous performance improvements and capability enhancements

Robust security measures

Security is paramount in AI operations. The Moveworks Reasoning Engine:

  • Operates within a robust security framework
  • Adheres to industry-best practices
  • Holds multiple certifications (ISO 27001, 27017, 27018, 27701, SOC 2 Type 2, and CSA STAR Certification Level 2)
  • Ensures all data processing and AI operations meet or exceed global standards for confidentiality, integrity, and availability

By combining these advanced features with its core understanding-planning-executing process, the Moveworks Reasoning Engine offers a powerful, secure, and adaptable solution for enterprise AI needs.

The Moveworks Reasoning Engine is purpose built to drive productivity

The incorporation of the Moveworks Reasoning Engine into our Copilot platform represents a significant advancement in AI-powered employee support. This sophisticated system is reshaping how organizations approach problem-solving and task management in the workplace.

Key capabilities

The Reasoning Engine enhances Moveworks' Copilot in several crucial ways:

  1. Contextual accuracy: It provides more precise and relevant responses to user queries by considering the full context of each interaction.

  2. Autonomous handling of complex tasks: The engine can manage multi-step tasks independently, reducing the need for human intervention.

  3. Adaptive problem-solving: It can adjust to new situations and challenges without requiring explicit reprogramming.

  4. Human-like support at machine speed: The Copilot offers support comparable to human agents but with significantly faster response times and 24/7 availability.

Real-world applications

Reasoning engines like ours find applications across various business functions:

  • IT support: Automating troubleshooting and resolving common technical issues.
  • HR operations: Streamlining onboarding processes and answering policy questions.
  • Customer service: Providing quick, accurate responses to customer inquiries.
  • Data analysis: Assisting in interpreting complex data sets and generating insights.

Why CIOs should take notice

For Chief Information Officers, the value proposition of a reasoning engine like Moveworks' is compelling. It represents more than just automation—it's an evolving intelligence that grows with your enterprise.

Key benefits for CIOs

  1. Efficiency at scale:

  • Automates tasks that typically require manual oversight
  • Frees up IT teams to focus on strategic, high-value work
  1. Reduced support workload:

  • Lowers ticket volume for support teams
  • Enables IT staff to tackle more complex, innovative projects
  1. Enhanced data accuracy:

  • Prioritizes and cross-references information
  • Ensures decisions are based on the most reliable and up-to-date data
  1. Streamlined operations:

  • Acts across multiple departments and systems
  • Reduces operational bottlenecks and improves overall efficiency
  1. Continuous adaptation:

  • Learns and evolves with your enterprise
  • Stays ahead of changing needs without constant reprogramming

In essence, a reasoning engine like ours offers CIOs an opportunity to significantly upgrade their enterprise AI capabilities. It enables intelligent, autonomous problem-solving that improves with each interaction, driving continuous improvements in operational efficiency and employee satisfaction.

By implementing such a system, CIOs can position their organizations at the forefront of AI-driven innovation, ensuring they're well-equipped to meet the evolving challenges of modern business environments.

Better reasoning, better employee support

Through the Moveworks Reasoning Engine, Copilot can do more than deliver answers; it can strategize, learn, and adapt, much like a human agent. This agentic approach ensures that employees get thoughtful, tailored responses, no matter the complexity of their request. It’s the ultimate blend of AI-driven automation and human-like reasoning, built to save time and elevate productivity.

The Moveworks Copilot puts powerful AI directly at the fingertips of employees to aid them in their day to day. The Reasoning Engine can perform the same diligence as a human agent when it comes to comprehending a user's issue, developing a plan to fix it, and executing the appropriate actions to do so with the same degrees of success — only AI can do it faster, for cheaper, and around the clock. 


See the Moveworks Reasoning Engine in action. Sign up for a personalized demo here.

Table of contents


Subscribe to our Insights blog