Tuesday, May 12, 2015

Best Agents for a Self Driving Process

Knowing which agents to leverage in your new digital processes, that emerge perfectly to support innovative and evolving business models, will be a key evolving skill. Since this is an evolving science that is emerging itself, I thought I would take a shot at categorizing the typical new agents that will be used in the new digital world or the use of older agents in the new digital context. In order to design a jump start emerging process, inviting the right agents to the mix will be important. The rating system that I chose is in the figures below with two building block blog posts to review for details, if desired. I have identified 15 agent types to describe and rate, but these will probably evolve over time so use this a starting point.



















http://jimsinur.blogspot.com/2015/05/self-driving-processes-are-built-to.html
http://jimsinur.blogspot.com/2014/10/iot-what-does-your-thing-look-like.html





























The ability to swarm means that the agent has the ability to collaborate with other agents, especially humans, while seeking common goals shared by the other agents. The more intelligent an agent is will determine the effectiveness of the emerging swarm to achieve the dynamic goals. While some of the 15 agent types can be either human or not, the more specialized the outcomes, the more likely that software or devices are the resources that swarm with nano second speed. With more machine intelligence emerging in the digital world, this trade off will change over time. The first two agents for the figure above will be assumed to be humans for now.

Managers are the people who look laterally at interactions between resources and those folks who set and change the goals, policies and constraints. They guide the resources and coach the workers to get better outcomes and to grow in capability. In the digital world the goal changes and mixes change much more frequently in these self driving processes

Workers are the people who are generally "heads down" and have specific actions and behaviors that leverage their specific knowledge and skills for the good of the overall desired results targeted and coached by the managers. In the digital world, some of the workers will have enhanced knowledge with the help of other agents to become more generalized in some instances.

Negotiators are most likely a piece of software that decides which agent actually does a portion of the emerging/swarming process by weighing the bids by the intelligent agents and knowing the specific capabilities of each of the non-bidding agents. While some of this can be pre-planned in snippets of fixed actions and sequences, much will be dynamic in the new digital world. In fact there could be several collaborating negotiators working to the same goals sets. This is particularly true in the operation of end to end processes like a customer pleasing process, a value or supply chain.

Personal Assistants are most likely a highly interactive piece of software that knows the human resource it is supporting very well and can deal with requests in a friendly manner. In some cases they can predict the patterns of activity ahead of time and anticipate an individuals need before they request them. In most cases they can sense the persona and adapt to their needs in a real time fashion with the utmost of grace. These will be a fast growing segment of agents in the new digital world.

Cognitive Agents are the new class of software agents that are self learning artificial intelligent in nature. The learning can range from complex knowledge / pattern searches to real deep learning, but the aim is to supercharge the humans and other agents that are involved with self driving processes need in the new digital world.

Agile Process Snippets are basically sets of actions or tasks that are sequenced in an agile or externally controlled fashion. These are in contrast to dedicated snippets that have no ability to dynamically change other than using them in a different context. These agile snippets usually have explicit rules or are dependent on seeking external data parameters for the behaviors of the actions.

Multi-Functional Robots are general purpose robots that can interact with various other agents and can adapt to requests. These robots tend to interface with humans better and quite often are guided by machine learning of some sort.

Smart Application Program Interfaces are leveraged to use multiple and differing dedicated APIs in aggregations. These composed and leveraged software applications will have to be smart to hide underlying differences, transformations and communication protocols.

Smart Composite Applications are software applications built by combining multiple existing functions into new applications. This can be compared to mashups for software aggregation. In the new digital world existing components will have to be combined and used in new and innovative ways.

Complex Events are involved with processing that combines data from multiple sources to infer patterns that suggest complicated circumstances. The goal of complex event processing is to identify meaningful events (such as opportunities or threats) and respond to them as quickly as possible. In the new digital events of interest will emerge beyond the expected event patterns and must be dealt with quickly.

Dedicated Services / Snippets are individual actions represented by the fixed interaction of software and human actors. The sequence and outcomes are unchanging and can be easily modeled and only changed with new versions of the service or process sequence. They contain predictable functionality and do not vary in behavior. They are often leveraged from a portfolio of existing services in the new digital world.

Specialized Bots are mechanical or virtual artificial agents, usually an electro-mechanical machine that is guided by a computer program or electronic circuitry. They tend to be singular in focus and function.

Controllers / Clusters are the aggregation of multiple event sensors for local pattern sensing, decisions and actions in physical devices. They can act independently or in concert with an emerging process. in the digital world, more actions will be pushed closer to the event sensing and notification will be sent after the fact unless the conditions warrant independent intervention.

Dedicated Application Program Interfaces are a set of routines, protocols and tools for building and leveraging software applications or pieces there of. The API specifies how the software components should interact with software components, particularly user interface components. In the new digital world, agents will have to publish their set of interfaces to cooperate with other agents.

Sensors are generally devices that sense (that is, to detect) some characteristic of its environs. It detects events or changes in conditions or quantities and provides a corresponding output of the sensation. In the real time digital world that interfaces with the internet of things (IoT), these event sensors are critical for both physical and logical condition changes.

Net; Net:

The best agent for a self driving  and emergent process will depend on the situation, but identifying the types and use of these agents in the new digital world is our challenge.  While the figure of the 15 agent types is not a periodic table for agents at this point in time, it is helpful for process designers and participants.  It helps to understand the potential swarming participants and their roles in self driving processes.