Johnny Knoblauch

In 2011, the CEO, Michel RACAT, hired me as a new CTO to help them scale their solution. At the beginning, it wasn't easy; the product wasn't stable at all. Only one person could maintain it: Jean-Baptiste GABIN, my predecessor. He did a great job and developed the first version of the BeezUP product himself. The existing technical team was an offshore team and not dedicated enough to help me, so I fired them all. My predecessor agreed to give me a month, even though he was in terrible shape. Michel trusted us and we passed on his product knowledge so that he could leave the company with his head held high. I built up my technical team to maintain the existing system at the same time. Then I added new features... In 2012, we first developed our own billing/contracting system to automate the subscription process and offer a true SaaS product to our customers. In 2013, we migrated the BeezUP system from Ikoula to Microsoft Azure, and I developed our SQL Job scheduler on Azure. In 2014-2015, BeezUP has a generic product/offer publication system & a generic marketplace order management system. We integrated Amazon, eBay, Mirakl, etc. In 2016-2017, the whole BeezUP system is a RESTful Web API: BeezUP Merchant API Specification (https://api-docs.beezup.com) and uses Azure API Management and SwaggerHub. New merchant BO: https://go2.beezup.com is a Single Page Application. We rebuild our catalog importation & channel management system using CQRS/ES, DDD, and Actor Model with Azure Service Fabric. In 2019, BeezUP allowed marketplaces to be compatible with BeezUP by respecting this API contract, BeezUP Marketplace Adapter API Specification (https://mkpadpt-api-docs.beezup.com/). In 2020, the whole BeezUP system is fully continuously delivered on production automatically with nuget reference auto-update in place. In 2021, BeezUP was the first e-commerce feed management with a CMS that centralizes the discussion/messages with the merchants' buyers coming from the marketplaces.

I realized that even if my team could deliver the features with:

- great talent

- good team communication skills

- great architecture and technical stack

- great code quality

- great test quality and coverage

- within budget

- on schedule

It wasn't enough, our customers/users weren't using it, and even those who were putting a lot of pressure on us to deliver as many features as possible in a short period weren't using it. And why? So I made my product management skills by reading the book User Story Mapping by Jeff Patton (https://www.amazon.com/dp/1491904909/ ref=cm_sw_r_as_gl_api_glt_fabc_943R88XEDXR0C93EARN1? linkCode=ml1&tag=jefpatass-20) and taking the training course "Passionate Product Leadership Workshop" with him. Everything became clearer. My technical team wasn't a product team with a product culture. My technical team was a feature factory rather than a team that resolves customers' problems. So I renamed my technical team to a product team. I have put in place 2 teams inside the product team:

- Opportunity team: anyone in the company can submit & evaluate opportunities with the product team

- Discovery team: Define what to build with a cross-functional team to know if it's Valuable, Feasible, and Usable. Now I'm working hard to get the right product management cycle going: Sense, Focus, Discovery, and Delivery. Instead of a roadmap, we now use an opportunity backlog where we do a business plan for each opportunity by simply using the opportunity canvas from Jeff Patton. We have removed the role of Product Manager because everyone is PM now. Now I'm studying Teresa Torres Opportunity Solution Tree to improve visibility, and easily share with stakeholders all the paths we can take.

In 2017, BeezUP was bought by Cdiscount to create the IMN Platform (imnnow.com). We implemented code generation based on a DSL derived from swagger specifications to improve our productivity. With this, we can generate:

- the structure of all our back-end microservices

- the data structure like events, commands, messages, and models

- the client side in C# and TypeScript

- the documentation

In 2018, the IMN platform was fully developed. The technical stack has been improved:

- Azure Kubernetes Service instead of Azure Service Fabric

- Actor Model Proto Actor instead of Azure Service Fabric

- GraphQL API instead of REST API where possible

- IMN Marketplace API Specification provided for the marketplaces (https://mkpadpt-api-docs.beezup.com/)

- IMN Gateway API Specification to be implemented by the marketplaces (https://gateway-api-docs.imn.io/)

- This platform has 3 personas, so we have developed 3 Single Page Applications:

- The merchant user: https://go.imn.io/ using IMN Merchant API Specification (https://api-docs.imn.io/)

- The marketplace agent: one back office per marketplace that uses a GraphQL API

- The system agent: https://system.imn.io/ uses a GraphQL API

Everything is completely secured by Azure Active Directory and Azure AD B2C. The IMN platform has been developed with a zero-bug policy and is fully event-sourced. It's a pure clean code system. In 2019, the IMN platform was successfully launched, reaching its 2-year goals in 2 months.