Automation ​

The course talked a lot about automating operations - in the form of habits, rituals, or at least checklists. Such automation allows avoiding problems when performing constant repetitive actions. For example, even highly qualified agents can avoid situations like "attention involuntarily distracted by a colleague's question or something else, and then numbers were mixed up in the report, and then forgot to check."

Automation as the execution of instructions is related to S1. It allows quickly performing actions and using skills honed to unconscious competence. Therefore, we want to automate repetitive actions for a repeating role to the maximum.

The simplest automation is achieved through symbiosis with the exocortex. Gadgets can remind you of a meeting, Pomodoro - help maintain focus on tasks. Some functions can be delegated to the exocortex, for example, you can automate report collection using scripts.

Slightly more complex automation is achieved by learning proper behavior models and repeatedly practicing them, gaining "flight hours." For example, you study the technique of late running and start practicing proper running, constantly training and correcting errors in technique execution. Then sooner or later the skill will be internalized, and you will automatically take the correct postures while running.

Learning theories suggest that agent actions automation is non-linear. First, the first success can be obtained quite quickly, for example, it is possible to identify a role and key objects for this role. This can be followed by success consolidation: more roles and key objects for these roles can be identified. Next, the skill can be transferred to another context: identify the roles of several colleagues in a meeting, plan the creation of a work product per role.

Eventually, the agent will encounter problems. For example, it may be found that the role cannot be named, and instead of a role, an attempt is made to name a position, for example, "leader." Or that the role name does not explain the function the role performer will fulfill. For example, what exactly will a "doctor" treat? Or that the role name affects how it is performed: "analysts" only "analyze and recommend," while "engineers" can suggest "engineering solutions." This is where the "learning dead zone" begins.

Instead of swiftly transferring the skill everywhere, you will have to slow down and first figure out how to identify and name roles. You will also have to correct errors that will inevitably arise: how not to confuse a position that indicates authority over resources and changes less frequently with a role, which may not be related to a position at all (a "debater" role in conversation) and refers to the agent's demonstrated behavior?

When the number of errors decreases, the skill can be transferred to new usage situations. Initially simpler situations, where there is time to think before identifying roles: for example, transferring to a situation of scheduling by roles when preparing for a meeting. Then slightly more complex situations, for example, learning to identify actually played roles in planned meetings, cross-referencing with the description made before the meeting: who was expected to play which roles? What was done in reality?

After that, you will be able to identify roles "on the fly" much faster, directly in a conversation or during a random encounter. And if you can identify roles accurately and instantly, even if you are distracted, it means that the skill has truly been automated.

But automation does not happen instantly. Usually, in a skill, individual small operations that need to become automatic can be identified, even if the skill is relatively simple, like using Pomodoro. The Pomodoro application needs to attract your attention - for this, you will have to design the path of your attention. For example, place the application icon on the desktop/main screen so that you will definitely look at it and launch it. A task for which you start Pomodoro must be defined, and it is preferable for the work product for it to be understood so that you associate Pomodoro with productive work rather than just turning it on. After the end of the Pomodoro/session, you will need to take a break to maintain performance until the end of the day. And the execution of these small operations should be automated, so that you can talk about installing the Pomodoro practice within yourself.

Automation will allow you not to engage S2 needlessly. However, the automation of operations does not mean that you can completely turn off your brain. First of all, during the automation process, you won't be able to do that: learning or writing scripts will still require you to think. Secondly, even when you have a working automation, you will have to face situations where choices are needed. For example, if you set a rule "do not be distracted when Pomodoro is running," does that mean that you cannot stop working when the fire alarm goes off and are strictly required to wait for the Pomodoro to finish? No, of course not. If you are a salesperson and you have written scripts, you will eventually encounter a situation where you will need to deviate from the script to attract a good client. Extreme approaches like "do not automate work at all" or "automate until your brain turns off" are not needed. You will have to train the skill of distinguishing situations where working with a script is possible (most of them if the script is good) and those that require thinking. The skill of differentiation is complex, and before training it, you need to make sure that the agent is capable of thinking/reflecting, not just following instructions. If not trained to think, you will first have to learn to think, and then differentiate situations. The process is not quick, but there is no other way to get a highly qualified role player.