We need an Autonomous Agent Oriented Architecture

In order to encounter new developments in big data, internet of things and FinTech, organizations need to adapt a new architecture pattern: the autonomous agent oriented architecture (AAOA). The older concept of Service Oriented Architecture (SOA) turns out not to be sufficient, too cumbersome, not flexible enough and inappropriate to handle the increasing streams of data. With SOA we are drowning in the data lake.

The Autonomous Agent Oriented Architecture is characterized by an extreme degree of distribution, complete autonomy of parts, and emergent control. It supports the API economy and practice of micro services, eco-systems without central control. It is a reaction to the current imperium view that characterizes the Service Oriented Architecture (SOA).

The Autonomous Agent Oriented Architecture is an architecture pattern that focusses on autonomous units, behaving as autonomous agents, that can live and survive in a lake of data without the need for a central control system. Agents can be organization units, application units (apps) or infrastructural units (agents). Autonomous Agent Oriented Architecture can be applied on Business Architecture, Information Architecture, Application Architecture, Technology Architecture. The main element is the Agent, which is an autonomous unit that is capable of operating without the need for a central control structure. The only context that an agent needs is a physically based communication network to which other agents and systems are connected, and data that is communicated between them.

The anatomy of an agent

An agent, in order to be autonomous, needs to have:

  • senses - interfaces that intercept messages or signals (data)
  • cognition (business logic) to interpret sensed data
  • decision knowledge (business logic) to make appropriate decisions. This may contain artificial intelligence and machine learning capabilities.
  • actionable part - interfaces that are capable of doing something that has impact to the outer world, or just sending data to whoever picks them up.


Let’s look at an example. When our refrigerator is going to contain its own computer to control the fridge, and to communicate the status of its contents, it is an agent. It can look in its own contents, and it can communicate with other agents. Another agent may be the weather station, who is sensing the outside and inside temperature and air pressure and sending messages that are received by the fridge. Other agents may be the smart thermostat and your car in the garage. Our houses are getting full of these autonomous agents, communicating with each other without a central control. Our domestic architecture will be an agent oriented architecture: we can bring in and throw away all kinds of autonomous gadgets, without corrupting the whole structure and without impact on the whole architecture. An Autonomous Agent Oriented Architecture is invulnerable to change: it embraces change and accepts the lack of central control.

Why Autonomous Agent Architecture for organizations

Wouldn’t that be the ideal architecture for large corporations like banks, insurances, telcos and public services? 

Consider a bank, not as a hierarchical construction of business lines, but as a network of business units. Consider an insurance company, not as a monolithic organization, but as a set of autonomous agents that each take their own responsibility. 

Consider the application landscape of public services not as one block of services and applications that are controlled by a ESB or BPM layer, but as application agents that act autonomously. In many cases I am convinced, this reflects reality more than we as architect tend to think. 

The application landscape of a large corporate organization would highly benefit from the recognition that it actually is a landscape of agents; we only need to enable them to act autonomously. The organization landscape of e.g. the Dutch government is actually a landscape of autonomous organizations, some of them even named “agency”; why then an enterprise architecture that brings in central control of an ESB and pictures the organizations as an “imperium”?

It is time to recognize the autonomy of organizations, of applications, of data, in our enterprise architectures. It is time to survive our increasing data lake, otherwise we are drowning. It is time for an Autonomous Agent Oriented Architecture.

Weergaven: 529


Je moet lid zijn van Via Nova Architectura om reacties te kunnen toevoegen!

Wordt lid van Via Nova Architectura



© 2020   Gemaakt door Stichting Digital Architecture.   Verzorgd door

Banners  |  Een probleem rapporteren?  |  Algemene voorwaarden