Monday, October 1, 2018

Working on sharing my notes

I started to develop this application in Angular 6 deployed in as a page app. This is fun to do. I decided to do this content as I do swap a lot of different technologies depending on the customer engagements or assets to develop.  Maybe it can be useful for other.

ODM Product Recommendation for customer churn

We presented last week an important asset we have developed to integrate a predictive scoring with business rules and chatbot for doing product recommendation. This open source content, visible at, and includes how to use IBM Decision Composer to start modeling decision with Decision Model Notation, and move the content to IBM Operational Decision Management. Then we modified the data model and implement product recommendation rules. We highlight how to deploy ODM On IBM Cloud Private the Kubernetes platform, how to implement ODM simulation to find the best rules implementation to improve business impacts, and how to leverage churn scoring to propose discount.

This asset explains the integration with a chatbot developed with Watson Assistant, and a larger solution, named "green compute", which addresses a solution implementation and analytics reference architecture. The use case is detailed in this repository with a detailed demo script and all the code to run it, we explain how we built the machine learning with python spark API leveraging customer records, chat transcripts, and marketing campaigns. 

This is a very interesting contribution that we are improving over time.

The contributors are Guilhem Molines,  Mathew Voss, Joseph Luc Correa, Jordan AcockZach Silverstein and myself.

Thursday, May 10, 2018

Hybrid Cloud Integration Solution Implementation

Hybrid cloud integration represents interesting challenges on how to protect existing SOA services with new cloud-native web applications and microservices. Co-existence is a must on today IT landscape. I'm leading an effort at IBM to implement a complete hybrid solution where architects and developers will be able to learn best practices around:

  • how to develop a SOAP app in Java using JPA, JAXWS deployed on WebSphere Liberty
  • how to develop gateway message flow within IBM Integration Bus
  • how to define API product with API Connect, and use secure communication with TLS for backend APIs
  • how to set up secure connection from IBM Cloud public to on-premise service using IBM Secure Gateway
  • how to develop a Single Page Application with Angular 5 using a Test Drive Development approach with nodejs/expressjs back end
  • how to secure the web app with passport
  • how to access existing LDAP service for user authentication
  • how to perform CI/CD in the hybrid world
  • how to monitor all those components using Application Performance Monitoring
  • how to deploy most of the components of the solution to IBM Cloud Private
  • How to call BPM process from Watson Conversation (orchestration), and how to integrate chat user interface connected to Watson Conversation into BPM coach.

Visit the top level Github repository from which we can navigate to the different content:

We are updating these repositories on a weekly basis and we welcome contribution and comments.

Tuesday, January 23, 2018

The new trends of IT development for business applications group three important needs:

  • how to develop and deploy applications, business processes and decisions faster than ever to operate at a massive scale.  Those applications include new capabilities like chatbot, cognitive integration and machine learning capabilities to improve human to machine interactions
  • developers want to use their own languages, components, APIs, and frameworks so they can develop quickly new business application, deploy. measure and learn from production execution. They need to get the job done, agile way, building by short iteration and incremental capabilities that bring quick business values to the line of business. They need to adopt lean startup approach, A/B testing and pivot if needed. 
  • Access to resources, compute capabilities, added value services, on demand, as a  service consumption with simple pricing, and test and run as they need to scale in and out. 
In term of business leads, digital transformation is the major vector for change: augment business process with AI capabilities, improve integration to existing systems, public , private APIs, optimize decision execution, improve human experience. 
I'm presenting next week in madrid how a technical implementation supporting those needs may look like by integrating API management, integration, Java based microservice, cognitive services, scoring service based on machine learning running on Sparks. All the content, how tos, code is open sourced and can be found here:

Friday, January 19, 2018

Our team just publicly published a white paper on how to use Watson Discovery on the IBM Cloud platform to create content collections and custom cognitive applications. It is related to the training tutorial I co-developed with colleagues at IBM but adds other best practices and business use cases. You can read this pdf here: IBM-Advantage-Paper-for-Cognitive-Discovery.pdf
Please give us feedback.

Wednesday, January 17, 2018

Complement Watson Conversation with Business Decision

Today I presented to IBM Reference Architecture team, with Guilhem Molines, IBM Operational Decision Management product architect, a new solution we are using during our consulting engagements that leverages the Watson conversation to gather customer's input and send to a decision service running in ODM. The solution is implemented as part of ( so every IBM customers or consultants can study it.
Integrating chatbot in business applications, is definitively a trend started two years ago, and a lot of companies are still implementing solutions today. As part of the continuous improvement and optimization of the business operations, the next step is to use the data gathered during the conversation and apply business rules on top of it to deliver next best action or best recommendations. This will limit the human intervention to complex cases only, offloading the workload to rule engine.
To play with our solution you need an IBM Cloud public account, and the code and artifacts we developed to be able to run this solution. The solution integrates with [Decision Composer].
Have fun with this new assets and feedbacks are welcome.

Wednesday, January 3, 2018

IBM Cognitive Reference Architecture

I am part of a group developing reference architecture and assets for IBM Cloud division. One of the major and most visited content is the Cognitive Reference Architecture

We are starting a set of Video on Cognitive Reference Architecture

And the implementation is progressing at

Feel free to contact me for enhancements and feedback.