The Four Fundamentals of Intelligent Automation

Using the John Boyd Decision model as a reference, Cortex Intelligent Automation Software encompasses the four fundamental elements of Intelligent Automation: Sensing, Analysing, Deciding, and Acting within our established SADA Model.

The SADA Model

Cortex Sada Model

  • SENSE, monitor, and decipher events, receive requests and understand situations.
  • ANALYSE situations using machine intelligence and critically manage all exceptions.
  • DECIDE using logic, rules, inference and neural nets to orchestrate an automated response
  • ACT, to apply the required automation, processes and functions based on the previous cycle

The Types of Automation

Automation can be easily categorised into types that build a hierarchy of automation ultimately resulting in a system.

Has common capabilities typically seen in mass produced technology applications. Its function is unchanging, for example, posting an SAP invoice entry.

Functional automation carries no context; it will always work the same way wherever and whenever it is used. For this reason, it is most found in off the shelf products.

Functional automation can be augmented with missing functionality by common task automation tools and techniques such as Scripting, Coding, RPA, and Runbook type tools.

Functional automation is all about reliable, repeatable actions.


Workflow automation automates the sequence of operations, defined as the work of a person or group with common skills, knowledge or resource access.

Many Workflow automation tools emanate from the business process management (BPM) sector, where workflows are defined in forms-based sequences of screens. More advanced BPM tools allow business rules to be applied to route the flow of work to various actors (people or systems) based on skill, knowledge, responsibility or resource access.

Workflow automation can be confusing for people to differentiate from the process. Simply, a workflow is about flowing work between different groups of people. Workflow automation allows people to analyse and decide what actions are taken at each stage. Workflow is a high-level abstraction of common steps for different processes and normally emerge for efficiency where the process is poorly defined.

Workflow automation is characterised by the fact that every action starts and ends with a person.

The main driver for Workflow automation is improved repeatability and the removal of management time for directing authorising activities.


Task automation contains actions, configurations and governance specific to the individual organisation.  It is highly repeatable within a fixed context.

Task automation is a build once and deploy approach much like functional automation – but only for that specific organisation.

From a static input data and start state, Task automation executes the appropriate sequence of Tasks to produce success or fail outcome.

If a Function requires localisation, or configuration to work, it is known as a Sub-Task. Most often used to provide external system interfaces such as Servicenow, or Salesforce.

Task automation automates an activity in a single skill, knowledge or resource domain. It is dependant only on the initial conditions for a success or fail outcome.  Task automation is unchanging during execution which by its nature has a short running time.

Task automation is most effective in a workflow or case-based environment. As it will increase the quality and scalability of the output but will not have a significant impact on the mean time to respond, or the headcount.


Processes automation is sensitive to its context.

Process automation takes a transaction with a set of input data, and start state, through a sequence of decisions and tasks to achieve an outcome at an end state. The resulting data may or may not define the end state.

Process automation has multiple permutations chosen depending upon the details of the transaction, and the internal results of Tasks.

Process automation must be stateful so that analysis and decision can be taken, and exceptions generated.

Process automation can be single domain, single system, or cross domain, multi-system, and any combination. It is still Process automation, not “Orchestration”.

Process automation, in combination with Orchestration, achieves significant increases in velocity of operations as tasks are executed and sequenced at machine speed. Significant headcount reduction will result from the exception only environment.


Services are sets of activities that produce customer outcomes such as provisioning a network.

Service Orchestration directs the appropriate set of actions: Processes, Functions, Tasks, Workflows, and Services to achieve an optimum outcome.  This can be to support an objective, deliver an outcome, or maintain a goal.

For example: To maintain the performance of a software application. Technical Service / Resource Orchestration is the coordination of tasks and technology systems and services to achieve this.

Service Orchestration tools are hierarchical in nature and are Analysis and Decision driven to align with operational structure. This differs from simple orchestration tools which have two dimensions, decision and action.

Service Orchestration in conjunction with Process and Task Automation can fundamentally transform the way in which businesses operate. Successful service orchestration will deliver cost saving, headcount reduction, quality improvement and velocity (MTTR) transformation.


Automation and Associated Technology

Cognitive Computing

Methods and Algorithms that allow machines to recognise and categorise things. For example, Natural Language Processing allows computers to recognise speech, and Imaging technology allows machines to identify images.

Artificial Intelligence

Machines emulating human intelligence which currently don’t exist in any real-world implementation. Intelligent systems with 3rd and 4th level automations can exhibit characteristics that are often confused with Artificial intelligence. This is not an emulation of how the human brain works.


RPA is the execution of a task through machine emulation of human user access to applications via human interface devices such as Windows, Keyboard and Mouse controls, or Command-Line Interfaces. This is often done to capture data from an application when an API or database access is unknown, unusable, inaccessible, or not available.

Machine Learning

A machine is able to modify its own programming to improve or optimise outcomes.