Organizational Ontology: Driving aiOla’s Actions

aiOla’s motto, “speak productivity into action,” highlights the need to consider spoken reports of organizations as an integral part of complex workflows. It’s not enough to transcribe speech into text with high accuracy. Text transcriptions are still unstructured, and therefore, not always useful for large organizations. 

Even aiOla’s natural language understanding (NLU) models that convert text into structured events aren’t enough to drive maximum productivity from the data since some events give a partial view of the overall workflow they’re part of. The last missing piece in the drive for productivity is the creation of organizational ontology, which makes it easier to define specific business logic policies for various structured reported events. 

In this blog post, we’ll describe the architecture of aiOla’s ontology that’s customized for each of our customers and look at how it can be used as a means to benefit from aiOla’s technology. 

Examples of business logic policies

Over the years, we saw a few repeated examples of business logic, requested by customers, that helped us implement aiOla’s “Speech to Action” technology. In this section, we’ll review some of these examples and look at how they can be adapted for other use cases:

It depends on the weather

Different weather conditions require different checks and policies. Consider the space shuttle Challenger, which exploded during liftoff due to the cold weather that affected some of its seals. One of our delivery logistics customers wanted to apply different air pressure standards in different weather conditions, such as in cold or snowy conditions. To achieve this, the customer provided the exact location of each facility, which allowed aiOla to check weather conditions in each location and notify the safety inspectors of risks when they measured the trucks’ air pressure before going on the road. 

You need a valid certification

As technology-related jobs are constantly evolving, the need for certification for new tools and new employees is only growing. It’s becoming increasingly difficult to monitor the actions of every employee and verify that they have a valid and suitable certification. One of our customers in the food safety industry wanted to restrict junior and inexperienced inspectors from reporting issues on one of their newer food processing machines before they passed a dedicated training program. To achieve this, the customer provided a list of employees and their certifications status. When employees spoke with aiOla to inspect these new pieces of equipment, our technology notified them whether to proceed or wait for a certified inspector.

3 strikes and you’re out

One of aiOla’s large retail customers wanted to verify that the temperature of the display refrigerators and freezers wasn’t deviating from the standard range for more than two or three consecutive reads. Before using aiOla, issues might get reported, but the displays were still in use unless the issues were repeated too often or for too long. However, it was difficult to gauge what was considered “too often” or “too long” since each unit was different and readings were based on the type of food in that unit. To achieve more flexibility, which was critical for the productivity of its operation, the retailer provided a list of each unit along with its unique settings. aiOla used this information to alert the retailer when temperature readings have been off three times in a row so that a unit could be reported to a technician. 

These are only a few of the many business logic examples that large companies need to improve complex workflows and achieve higher levels of productivity. In the past, the implementation of such business logic policies into data systems was slow and tedious. However, with the flexibility of the natural language technology we have today, the power of aiOla’s artificial intelligence models, and the ability to capture organizational ontology, it’s now quick and easy to define new use cases with complex business rules and deploy them at scale across organizations. 

Ontology Architecture

When building our ontology architecture, we faced a challenge: How can we build a flexible ontology for every large customer while still catering to strict and individual security, performance, scalability, and availability requirements? 

aiOla’s experienced team has been building enterprise-grade cloud systems for many years, and while challenging, the ontology architecture proved to be a great example of an elegant solution. By relying on a combination of different technologies, like the ones we’ll see below, aiOla was able to address unique security, performance, scalability, and availability needs that could be adapted to different use cases. 

Let’s take a more detailed look at the ontology module’s overall architecture and the main considerations in its design. 

The aiOla Ontology architecture is a combination of three main pillars:

  • Hierarchical data store – AWS Cloud Directory: Optimized data store for hierarchical (tree) data, which fit large and complex org charts and similar enterprise structures. Here’s an example of an ontology diagram:

  • GraphQL API – AWS AppSync: Flexible API optimized for traversal of multi-level hierarchies with a Lambda function as a data resolver in the cloud directory service. For example, the following query is used for multiple levels of the ontology:
query getRootOrganization {
 organization {
   regions {
     managers {
       employees {
  • Amazon Verified Permissions (AVP): Secured and optimized service to check access control and similar permissions based on flexible and complex policies in a policy store. For example, below is an example of a policy in AVP that allows only employees of each store to inspect that store and not others.
permit (
   action in [Ontology::CustomerX::Action::"InspectStore"],
{ == resource.store_number

Organizational Ontology

When a new ontology is created for a specific customer as part of the onboarding process with aiOla, a business analyst uses a dedicated combination of the above services and their schema. The three separate resources provide the needed tenant isolation for enterprise systems and prevent accidental data leaks and other security risks. By using AWS, these services also provide a high standard of security, high availability, and scalability to support almost unlimited ontology sizes and complexities. 

Wrapping Up

Adapting AI to enterprise customers is a challenge that aiOla is tackling head-on. The end-to-end solution that can be delivered on top of the aiOla platform takes unstructured, highly jargonized speech reporting of employees in the field and drives them through complex workflows with fine-grained policies to help organizations do more with less while still maintaining higher standards of safety and compliance. The ontology module described in this post provides unparalleled control for organizations to harness the power of AI in the most challenging environments and workflows.


Guy Ernest
Guy Ernest
Guy Ernest, the Chief Technology Officer at aiOla, is a seasoned leader in data, machine learning, AI engineering, and architecture. His experience spans working with leading enterprise clients across various industries such as CPG, airlines, telecom, media, financial services, and retail, focusing on developing and managing efficient Analytics systems. He formerly led Amazon's AI Solution Architecture worldwide, contributing to over 10 machine learning and AI service developments and launches within Amazon Web Services. Before Amazon, Guy held key roles in Analytics and Cloud Transformations at major firms like, Netflix, Supercell, Autodesk, Waze, Viber, IronSource, among others, across Israel, EMEA, and the USA.