The mountain guide - How we help customers to succeed in the wild west of machine learning


Yrieix Leprince
Oct 09, 2019 Column

The mountain guide - How we help customers to succeed in the wild west of machine learning

At craft ai, we're passionate about collaboratively driving success. The customer success team strives to be a trusted advisor to your internal team. The cooperative work starts by defining together what is possible to do with your data, and continues until the deployment in production of your new Explainable AI feature. Even after this successful launch we keep in touch to follow your growth and your needs.

looking for ideas

Our ultimate goal is to make your service more efficient than ever by deploying self learning AI and to help you to be more competitive!

If you’re collecting data, we can start right now, it is the only prerequisite. We come with an all-inclusive subscription model that scales with your success. This offer is threefold:

  • Access to our unique Operationalized Explainable AI technology (academically published),

  • A dedicated cloud infrastructure that scales as you grow,

  • Last but not least, a dedicated data scientist from our customer success team will guide and help you during your journey to a successfully deployed AI.

Let’s have a look at the key steps on your road to success with craft ai!

Let's get to know each other : every customer is unique!

step 1

The first step on your journey to success begins with a story, your story. Every customer is unique, so we have to understand your industry, your business, your objectives. No one knows your challenges and your data more than you do. This first step is of paramount importance to define the scope of the project. It gives us the information we need to help you benefit the most from our technology.

Once we have understood what is your goal, we can focus on the technical aspects. Our projects timeline looks like the following:

Prototype: your data set the foundation for success.

step 2

We believe your data set the foundation for success, and we are just here to help you to exploit it and enhance its value.

Our business model is simple. It only works with long-term partnership. Your success is mandatory. We might not have the final destination clearly defined yet, but we will work together in an agile process to increase your competitiveness.

To ensure our collaboration success, we define together a use case. Then we quickly implement a prototype based on your IT requirements and based on a sample of your data. It tests if your data can answer your need. This demonstration of our expertise has to be validated by your team.

This prototype is implemented in collaboration with your experts and our data scientists to use our algorithm in the best way possible. It has to answer your end user needs and validate what craft ai is capable of in production.

At this stage success metrics are precisely defined to estimate our algorithm relevance on your data. For example, in an energy provider company, a key performance indicator (KPI) is the number of buildings monitored by expert.

We now have a clear use-case, implemented and validated a prototype and defined success metrics. This period usually lasts 4 to 6 weeks, and then the project can go to the next step.

Pilot : integrate craft explainable AI features in your product.

step 3

After the prototype has been validated, we can begin to plug the craft ai API in your product. Our ongoing support now focuses on helping you integrating our API in your solution. Our customer Success members will be working with your developer team, whether they are in-house or outsourced, to enable craft ai features in the target host: dashboard, mobile app, web app, API, ...

As the figure shows it, your data stream is continuously updating our AI models with the latest records. Our customer success team helps you to implement a connector within your tool to enable Explainable AI features. This connector creates a link from your data stream to your craft ai dedicated cloud infrastructure. On your infrastructure/application side, another piece of code is added to retrieve and use craft ai explainable models.

Each dedicated cloud infrastructure is encapsulated: your data will never be used for other client in any way. Your data/models are your property at all time. Even if you decide to stop the cooperation, you will be able to retrieve all the generated models.

These steps aim at fully integrating our Explainable AI API in your production environment to test it on a reduced scope and/or for a reduced audience.

As we work in an agile way, we want to quickly have a first working version of this pilot to be able to track your specific KPIs. We iteratively enrich this version to make sure it meets your needs. While monitoring the KPIs the deployment scope is gradually increased.

At this moment you can evaluate in reality the expected ROI for the full scope

Production - Growing and optimizing your project

Once the pilot has been successfully validated on a reduced scope, it is time to launch it gradually over your whole system. Your team now takes full advantage of the Explainable AI. At this point we still continue to work alongside with you to measure, capture, and monitor your models.

This is where things start to get exciting! Driven by your insights and accomplishments, the work continues by developing and optimizing your solution through refinement or by growing new projects. Your business and priorities evolve, your data increases, your needs grow, so your product might need to be adjusted and scaled!

next opportunities

Whatever happens, the customer success team will be there to assist you on the road to success and competitiveness.


To sum up the key points:

  • From the beginning until the end, our customer success team helps you to maximize the profit of your data stream.
  • Your explainable AI based feature has been validated at each step by your expert team to provide to your end users exactly what they need.
  • As the continuous machine learning system is working since the prototyping step, and as it has been incrementally tested, launching it on the full scope is much safer.
  • We are always available to help you monitor and scale up your new eXplainable AI feature.

Did you like this article? Discover now two examples of successful collaboration with craft ai :

Interested? Contact us!

Article made with by craft ai.