Agile at Tesla -

I have been watching this video from Jon Justice in Agile Munich, and I would like to conserve some notes. I have seen some other parts of similar videos, so some notes comes from other sources.

Goal

How to build hardware in three hours.

Notes

Limit the number of people in a team (no more than 6 -15).
- Amazon: no more than 6 people.
- Tesla: no more than 15 people.
The importance of lines of communications.
They are mission driven companies.
Primary element of speed is keep people excited.
Removal of problems and hurdles.

How teams are launch

Teams are launched as opened space.
You, as employee has an application that tells you customer problems.
These problems are discovered using Machine Learning.
You are invited to a meeting where you will meet with others interested on the customer problem.
This individuals with their own background and opinions, suddenly are around a common goal.
These meetings take 3 hours.
How can remove pain in the next 10 minutes?
Switch teams happen when things are done.
12 hours shift is tough, but it’s an efficient way to do it.
Use of product backlog refinement, per shift (15 minutes duration).

Principles:

Reduce the customer pain to minimum.
Build cross knowledge teams right away.
Once you are hired, you are launch to this empowered environment.
Flat organizations (as much as possible).
People covering legal positions but changing.
No micromanagement.
Empirical control processes.
Automated tests as much as possible.
You don’t own nothing, it’s the group ownership that is working in a 12 hours shift.

Design and production are the same

Each part of the production line is responsible of delivering and improving their part of the line.
There is not an R&D department in other part of the organization.
If there is a heat-pump that is required to be improved, the production line detects this need, they look for the root cause, they look for alternatives, and they procure the new ones.
Every single car is different, but tested under the homologation tests.
Test driven development, operations and improvement: it’s called “Factory mode”.

All these approaches come with problems:

Focus on the right problems not always the right one.
Backlog management can explode in so many times, and you require to keep the bottlenecks identified continuously.

Digital twins

Every car has a digital twin car that are being tracked.
The around 2 million Tesla cars that are in the street are being tracked and Tesla is learning about these cars are being driven.
Customer feedback comes 24×7.
Tesla is using data driven analysis to understand the pain points.

Fair questions:

How are these feedbacks combined with long term vision of the products they are willing to manufacture?
How your maintenance teams learn about how to repair a car that probably is unique?
How are the legal roles being covered?

Tesla have software that is able to simulate physical conditions and enable teams to set goals and understand behaviors.

You as customer, can opt to be beta user, which means that you will be part of the community using beta versions of software on your car.

Bamboo line

If you’re going to inspect a product for quality, then whatever is framing that product must also be beautiful and flawless.
The vision is that it will emphasize any defects in cars coming off the production line.
Working environments need to be perfectly clean to ensure that no compromises are made on the product.
A clean environment is an environment which employees unconsciously respect, whereas, in a dirty environment, they are more likely to make careless errors and accept second best.
Factory lights are light and clear.
Use natural light as much as you can: employees exposed to natural light are more productive in the long term.