Workflow

Workflows are an extremely important part of Analytica: thanks to them the platform is proactive, takes initiatives and performs tasks autonomously.

They too, like the other components, are based on AI.

Trigger

Each Workflow must define a trigger, that is an event that makes it fire.
Triggers are divided into two types:
- temporal
- data driven

Temporal triggers allow the execution of a workflow at a certain time of the day, and its repetition over time. For example, we could define a trigger every day at 8:00, or every year on January 1st.
Thanks to these triggers I can therefore perform periodic tasks, such as data import, control of issued invoices, or updating of other data.

Data driven triggers, on the other hand, are activated when there is a change to the data present in the Knowledge Base. They differ, in turn, in that they can be activated when a generic data of a Knowledge Base type is modified, or on a specific data.

In the first case, it is possible to listen for creation, modification and deletion of any data of the specific Knowledge Base type. For example, it is possible to define a trigger for when a new invoice is inserted.

In the second case, however, it is possible to have greater granularity, and listen to changes to a specific subset of data. So, according to the previous example, it is possible to define a trigger for when a new invoice from a certain customer is inserted.
It is possible to define these filtering rules in the same way that Knowledge Base type Widgets are defined in the Dashboard.

Actions

The action is the description, through prompt, of what the Workflow should execute.
In some cases, the action can be very simple, and it could be the execution of a command like "Prepare a report of the invoices received this week and send it by email", up to much more complex actions.

If the set Trigger is of data driven type, the action will always have a report of the modified, inserted or deleted data available. In this way, it is possible to create actions that execute different things based on a state change, for example.

By exploiting the various Agents available, the execution of a Workflow could also generate the creation (or modification) of data within the Knowledge Base. This means that, indirectly, it is possible to chain the execution of different Workflows, even conditional ones, exploiting on the one hand the Agents and on the other the various possibilities of triggers.