AzureFeeds

This week we announced our agent loop, a groundbreaking new capability in Azure Logic Apps to build AI agents into your enterprise workflows. With agent loop, you can embed advanced AI decision-making directly into your processes – enabling your apps and automation to not just follow predefined steps, but to reason, adapt, and act autonomously towards goals.

Now, we are announcing the General Availability of our Azure Logic Apps Rules Engine. A deterministic rules engine runtime based on the RETE algorithm that allows in-memory execution, prioritization, and reevaluation of business rules in Azure Logic Apps.

The Azure Logic Apps Rules Engine is a decision management inference engine in Azure Logic Apps, which provides the capability for customers to build Standard workflows in Azure Logic Apps and integrate readable, declarative, and semantically rich rules that operate on multiple data sources. The native data sources available today for the Rules Engine are XML and .NET objects. These data sources are called “facts” and are used to construct rules from small building blocks of business logic or “rulesets”.

To create rules, you need the Rules Composer. It can be downloaded from the download center.

The Rules Engine can also interact with the data exchanged by all the connectors available for Standard logic app resources. This design pattern promotes code reuse, design simplicity, and business logic modularity. Our Rules engine uses a VSCode experience to create Logic Apps projects with Rules engine support. For more information on how to create projects with Rules Engine, visit here.

Now. What can I do with it?

In a world of AI that essentially follows a probabilistic approach, rules engines are vital because they provide consistency, clarity, and compliance across different business goals. When you use rules with a workflow in Azure Logic Apps, you can define the logic, constraints, and policies that govern how to process, validate, and exchange data across systems, while you avoid AI hallucinations. Rules also help you make sure that applications follow the regulations and standards of their respective industries and markets. By using a rules engine, you can manage and update your workflow’s business logic independently from the code and without having to alter your workflow. This approach helps you reduce the complexity and maintenance costs of your applications and increase their agility and scalability.

From a technical perspective, the Azure Logic Apps rules engine allows you to do forward chaining or forward reasoning, in other words to do a re-evaluation of rules triggered by changes in the facts because of a rule’s execution. This is one of those scenarios where rules engine is unique; instead of writing complex code or creating complex “state-machine” workflows, the logic apps rules engine conducts this task with an instruction called “Update”.

Getting started 

In the example below, I will show how to use the Logic Apps rules engine to ground an AI workflow loop. For this to scenario, I am adding a Rules Engine workflow, to an existing agent loop workflow, and use it to correct rates and provide a “cross-sell” recommendation.
First, I need to deploy the workflow from VSCode to Azure. As the rules engine currently only supports XML and .NET Framework objects, I create an XSD schema (using Copilot if you don’t have an existing one) and use it with a “Compose XML with schema” action to create the XML fact that is needed. To obtain the returned data, I am using the “Parse XML with schema” action as well.

After the logic app  was deployed, I added it as a tool in the Logic Apps agent workflow loop, with a Call workflow in this logic app. I then pass the values that I need for parameters for the Rules engine to work. And I leave the rules engine return values empty.

Then I updated my system prompt to indicate how I want the Rules engine to be used. The agent loop will find the right tool for the right job.

Once the system prompt has been updated, I proceed to run the workflow with a payload. I have highlighted in red in the Agent chat, the guardrails imposed by the Rules Engine. Those rules have been used to make sure that the AI responses fall within the internal compliance and cross-sell company criteria. Some of the business rules can have different priorities and might require re-calculation for accuracy. The Logic Apps Rules Engine takes care of it without coding or adding complex business logic through additional workflows.

Further adjustments to the rules using the Rules Composer will ground the agent’s results even more. 

What else can I do with it?

You can use a Rules Engine in any context. In fact, decision management that falls under Intelligent business processes automation is growing in customers who want to provide flexibility, governance and compliance with their cloud workloads.
Another well-known scenario is for BizTalk Migrations to Azure Logic Apps. For customers who have implemented the BizTalk BRE for decision management, content redirection, SWIFT or .NET framework.

Demo

Please watch the following short demo of this sample.

Contact Us

Have feedback or questions about the Rules Engine? We’d love to hear from you. 

Reply directly to this blog post or reach out to us through this form. Your input helps shape the future of Logic Apps and the rules engine.