Why developing a Bot

A bot can bring a lot of benefits for engineers. Here is a non-exhaustive list of the main families of topics addressed:

• Capitalization of knowledge
• Reduction of development lead time
• Verification of designs and inputs
• Decision support
• Data generation for AI learning
• Move to Cloud in-house application

Capitalization of Knowledge

Knowledge capitalization is a recurring theme for various reasons: how can knowledge be perpetuated in a company? How can engineers be given access to good business practices?

Reduction of Development Lead Time

The reduction of development lead time is also a recurring theme. The objective is to digitize an engineering process identically. In all the elementary tasks of such a process, there are 3 cases: the knowledge is well-structured, the knowledge is capitalized but poorly structured, the knowledge is oral. Depending on these cases, the required work will not be the same.

Verification of Designs and Inputs

The verification of designs and input data consists of building a bot to verify the robustness of data. For example, this type of bot can be used to verify the draft angles of 3D ribs in a CAD file, or to analyze business rules in an Excel file. The challenge on this type of subject is to be able to structure the knowledge to validate or not certain business quantities.

Decision Support

The decision support topics consist of helping engineers to generate a large number of possible solutions, and then to help them to find the best solution. One of the important points of these bots is the structuring of a score that makes sense for the business and allows the solutions to be ranked.

Data Generation for AI Learning

The topic family around data generation consists of helping supervised learning methods by proposing families of robust 2D and 3D solutions. The main issue is structuring sufficiently generic knowledge to have a realistic and heterogeneous response surface.

Move to Cloud in-house application

The move to cloud represents the set of subjects whose challenge is to transform a business application developed internally in a SaaS Cloud service. For example, the family of topics that lends itself to this exercise are the business Excel sheets containing a large number of VBA routines and whose objective is to offer a structuring of all the knowledge contained in the Excel sheet and VBA while offering a simplified Web interface for the end-user.

1. Value creation

One of the common points to put forward on all the previous subjects is the value brought by a bot. Indeed, it is important to address this question at the same time as defining the bot's outline. The development can become important, and the limits of the bot difficult to build, so an approach focused on the value created will help the developer.

More precisely, one of the difficulties is knowing how far to develop the knowledge bricks of the bot. A metric based on the value created is a good approach to help the developer to define its outline.

2. A few examples

Let's take, for example, a bot whose objective would be to automate the generation of welding points.

The first question is to know in which context we want to build this bot. Here are some possible contexts:

• Knowledge Capitalization: the objective is to structure a business knowledge dispersed through different tools.
• Reduction of Development Lead Time: the objective is to reproduce a business without trying to be exhaustive but only to automate design tasks.
• Moving to Cloud: the objective is to port a business application running locally on a PC.

The second question is to identify the value sought by the company in order to compare it with the cost of development. For example, if the goal is to reduce development lead time, it is important to quantify the time currently spent by engineers on this task and then estimate the time needed to develop the bot.

The answers to these two questions are crucial for the structuring of the bot. For instance, in our welding point generation example, different strategies for structuring the knowledge are possible:

• Optimization of the positioning of the welding points
• Search for the best choice of welding technology for fixed weld spot locations
• Minimization of the number of welding points with the objective of reducing the overall welding time