General Archives - Page 24 of 37

How DevOps’ is reaping the benefits of microservices!

Posted on 21/02/202015/06/2021 by Tushar

In the world of “one-up” IT, it seems like almost everyone is wanting to understand how to integrate microservices into their solution architecture. And rightly so. The benefits of using microservices are numerous and varied. Let’s examine some of these positives and consider if they can help you solve some of the problems that you are facing today.

Agile scalability

In a traditional Web application, you can author scaling rules such that when a certain request count is reached, additional instances of the entire Web application can be automatically spun up on newly (and dynamically) allocated Web servers. However, you do not have any control over which specific area of that Web application needs to be scaled. Do you need to scale the entire Web app, or just a part of it?

For instance, there could be a queued backup of HTTP requests waiting to post values into a payment system. Perhaps there is a rather extensive validation algorithm that needs to be processed before the payment can be finalized. This small yet often-accessed part of the Web app could thus become a bottleneck. In traditional scaling of a Web app, to scale up new Web servers would be dynamically provisioned, the entire Web app will be loaded onto those servers, and the process will continue.

A more efficient architecture design would be to scale up just the payment service on its own. Not only would that be quicker (since you are not provisioning new versions of the complete Web app), but it uses less resources (since you are only scaling a piece of the solution).

To further save in resources and time, add in a cloud provider to leverage its container technology to host microservices and other cloud services to help with innovation, scalability, provide resource optimization.

Focused development

The goal of “micro” service is to do one thing only but do it very well to meet the needs of the business. This typically means a small team, a finite and very focused development scope, and functionality to design and implement the microservice very well. Services should be “slim” and keenly targeted and no overlapping functionality with other supporting microservices. When this is accomplished, consumers of that service can thus focus on using that service and could care less about its implementation or how the microservice accomplishes its goals.

Keeping development focused at the microservice level gives teams the chance to experiment. It is very very easy for multiple versions of a microservice to co-exist for A/B testing. You can route certain percentages of requests to the different versions, assess /compare the results, and then make design decisions. This can be done multiple times per day if necessary. This ability to experiment is not possible with anything larger than a microservice.

When decomposing the monolith down to microservices, we are now free to implement each microservice in the most effective way. This may mean using a different version of Java or NodeJs than what the other microservices are using. Or you can go all the way in on polyglot development and allow teams to choose the specific technology (e.g. Java, .NET, Go, etc.) they want for their given service.

DevOps Integration

With all these independent teams developing non-overlapping and focused microservices, how to we integrate this all together into a viable solution? The DevOps model can serve as a facilitator. Both microservices and DevOps offer an agile model that is a key component of the microservices model. Well-designed microservices follow this model to assist in development, speed, and agility – yielding smaller and more frequent releases. Continuous Integration (CI) is all about integrating frequent releases and thus is a perfect platform for a microservices release model. This model encompasses shorter build, test, and deployment cycles that fuel the ability to quickly roll out new versions of a service.

Microservices bring additional productivity to DevOps by embracing a shared toolset, which can be used for both development and operations. That common toolset establishes shared terminology, as well as processes for requirements, dependencies, and problems. This encourages development and operations teams to work better with one another, allowing those entities to work jointly on a problem to successfully fix a build configuration or a build script.

DevOps and microservices work better when applied together. This is especially true when DevOps automation is added to the equation, ensuring you get the same process followed exactly each time through the CI/CD pipeline. Automation also cuts down significantly on the time to process the new code/build/test/deploy cycle.

Standardized Communication

Microservices communicate using common mechanisms, such as RPC (such as REST or SOAP), or messaging. This promotes easy interaction with them. With RPC, a service makes a synchronous request to another service, then waits for that called service to respond back. While it is a simpler programming model due to the logic of the caller continuing immediately upon return from the RPC call, it can also have blocking/waiting issues while calls are waiting to complete.

If asynchronous communication is required to avoid blocking calls, then messaging can be used. Here a message is “published” into a message broker. That broker takes the message and forwards it on towards “subscriber” receivers who have registered to be notified if this message is published. The publisher (calling code) can then return immediately after publishing the message to the broker and does not have to wait for the message to process.

Evolutionary Architecture

One of the big architectural benefits of microservices is how well microservices support the ability to implement an “evolutionary” architecture. This allows you to continually innovate and incrementally change without incurring any significant cost, risk and change to those services that are running on it.

The advantages of microservices

These are the primary positives of microservices. Together they form a very efficient and streamlined agile model for development. However, microservices are not a technology to blindly make everything better. There is overhead in developing them, and it is a more complex model in certain ways than a monolithic architecture. Due to this complexity, we must use the Agile development method, and DevOps automation.

AI and ML Advancement leading to new Application Emergement

Posted on 20/02/202017/06/2021 by Tushar

With most innovative technologies, at some point in the solution evolution there’s a tipping point from “this is cool” to “there are real business outcomes to gain.” We are now seeing that shift with artificial intelligence (AI) and machine learning (ML), especially in three major use case patterns:

1. Customer care

When it comes to customer care, many companies face a similar set of challenges: Customer service representatives (CSRs) typically need to view multiple windows on multiple screens to find answers to a caller’s questions. This results in calls taking a long time to handle, frustrated customers, lower CSR productivity and overall poor customer satisfaction.

AI and ML can help change these outcomes. One insurance customer implemented AI and ML to transform its contact center, which responds to approximately 5 million voice calls per year. It also introduced an intelligent voice agent to respond directly to 50-70% of all customer inquiries. The power of AI and ML extended and enhanced the customer experience and the productivity of CSRs, who are now able to deliver more accurate and consistent answers, at lower cost and greater speed. As a result, we have also seen reduced attrition, as the CSRs are happier in their roles and require less training time to become self-sufficient. Overall, by using AI and ML, the contact center is able to provide a high level of customer service, enhanced customer satisfaction and more efficient engagement.

In another example, a customer service center handling 60M service requests per year is implementing AI “smart chat” to transform customer service, drive better quality, improve consistency and lower costs.

The transformation of an operation as complex as a customer call center is a multi-step journey:

Organizing and visualizing the data required to support typical workflows is a critical early step and can include authenticating/validating the caller, providing easy access to the data from different back end systems, initiating transactions and closing out the call.
A second step adds natural language processing (NLP) to allow the CSR to ask more complex questions. NLP uses a trained ML model to determine the most likely answers from a collection of large, unstructured documents.
A third step includes self-service for external callers, using voice and natural language to respond directly to client inquiries without requiring interaction with a CSR.

While the steps themselves remain the same, the sequencing may vary to align to the specific outcomes the organization desires.

2. Human capital management

By applying cognitive capabilities to human resources, companies are able to transform their human capital management (HCM) processes.

One valuable use case improves the employee experience by leveraging intelligent natural language conversation services to integrate data in operational HR data systems, such as Workday, and in knowledge management systems that explain policies, procedures and guidelines. Benefits include improved employee engagement, higher productivity and lower costs.

Other potential HCM use cases that are ripe for AI and ML include recruiting, selection and on-boarding of candidates. AI and ML models are already being used to help analyze and match job descriptions and candidate resumes, and to recommend potential jobs for existing employees.

3. Process rethinking/reimagination

There are also multiple examples of leveraging AI and ML for rethinking and reimagining processes.

In one example, an insurance company reduced claim settlement time and costs by incorporating AI and ML to improve First Notice of Loss (FNOL) initiation and data collection, as well as claim classification/assignment processes.

Companies can improve the speed and accuracy of FNOL by enabling claimant self-service using natural language voice/chat to initiate a claim. They can integrate external data sources such as weather and employ a fraud scoring model to facilitate claim classification and corresponding assignment to the correct processing alternative. Processing alternatives can include:

Straight-through processing without human involvement
Sending appropriate claims to the legal department for potential early intervention to reduce lawsuits
Rejecting claims because the deductible is higher than the claim
Assigning claims to a human analyst/adjustor

Predictive modeling to triage injured worker claims has been developed and applied to help determine “return to work” outcomes, along with typical treatment plans for various types of injuries (NOT replacing doctors) and estimated treatment costs. This technique optimizes case manager resource assignment and helps injured workers get back to work, improving service quality and lowering costs.

Applying an ML model to extract information from unstructured documents and forms helps to provide insights from previously unanalyzed sources. This capability works by accelerating processing of information buried in those large unstructured documents and drives new insights, lowers costs and improves processing efficiency.

In a fourth example, Anteelo is automating the underwriting process for home inspection to lower costs and improve scalability and overall quality. This use case applies visual recognition to highlight details in home inspection photos and use those details to consult the underwriting guidelines and then augment the underwriting decision.

As you can see from these use cases, the ability to improve the customer experience and gain operational efficiencies through AI and ML are real and very achievable. What other use cases can you think of for your industry? I’d love to hear your thoughts!

Techniques of making Cloud Application Migrations simpler

Posted on 19/02/202015/06/2021 by Tushar

Many system integrators and cloud hosting providers claim to have their own unique process for cloud migrations. The reality is everyone’s migration process is just about the same, but that doesn’t mean they’re all created equal. What really matters is in the details of how the work gets done. Application transformation and migration requires skilled people, an understanding of your business and applications, deep experience in performing migrations, automated processes and specialized tools. Each of these capabilities can directly impact costs, timelines and ultimately the success of the migrations. Here are four ways to help ensure a fast and uneventful move:

1: Create a solid readiness roadmap.

The cloud readiness stage is often referred to as the business case and/or involve a total cost of ownership assessment, rapid discovery, advisory services, and more. However, most of these approaches fall short of the up-front analysis really needed to decide which applications to move, refactor or replace. When you embark on a migration to cloud, job one is to ensure you are ready to go. Diving right into a migration without a roadmap is typically a recipe for failure. If you don’t know what the business case is to migrate, then stop right there. You should come out of the readiness phase with a high-level plan and an initial set of migration sprints and a detailed roadmap on how to address the subsequent sprints.

2: Design a detailed migration plan.

This phase, also referred to as deep discovery, will help keep migration activities on schedule. All too often, lack of proper analysis and design leads to surprises in the migration phase that can have a domino effect across multiple applications. This phase should include a detailed analysis of the scope of applications and workloads to be migrated, sprint by sprint. This agile process shortens migration time and ensures faster time to value in the cloud. All unknowns are uncovered, and changes are incorporated in the roadmap to ensure anticipated business objectives will be achieved.

3: Set a strict cadence — and stick to it.

Often, most of the planning effort focuses on getting past that first migration sprint. This phase requires a great deal of planning and preparation across multiple stakeholders to keep the project on time and within budget. Scheduling all of the resources and tracking the tasks that lead up to D-day is no simple job. These activities must follow a standard cadence of planning steps to ensure effective and efficient migrations.

4: Automate as much of the migration as possible.

The way your organization performs the migration can also impact the project. In most cases, tools that automate the application transformation process, as well as the migration to cloud, can mean a huge savings in time and money. Unfortunately, there’s no off-the-shelf automation tool to ensure success. The migration phase requires people experienced in a range of tools and how to orchestrate the work.

One could argue that a fifth way to ease migration headaches is optimizing applications during the process. However, there are different schools of thought on whether to optimize applications and workloads before, during or after migration. I can’t say there is a definitive right or wrong answer, but most organizations prefer to optimize after migration. This allows for continuous tuning and optimization of the application moving forward.

It is important to note that steps in the migration process are not one-and-done activities. The best migration and transformation services are iterative and continuous.

You can’t determine readiness only once. The readiness plan needs to be reviewed on an ongoing basis. Don’t analyze and design migrations just once. Do it sprint by sprint, and plan each migration activity. Make sure you continuously collect and refresh data, refine and prioritize the backlog, and look for key lessons learned to ensure continuous improvement through the lifecycle of entire migration. Experience and planning can go a long way toward easing your next application migration.

How learning aids in the development of an agile workforce

Posted on 17/02/202015/06/2021 by Tushar

While forward-thinking companies are rapidly investing in their technology resources, they’re doing the same for their human resources.

That’s because these organizations know that the workforce of the future isn’t made up solely of robots. Instead, the workforce of the future is an “augmented” one, where humans and AI work together to drive greater efficiency, innovation, and business success.

Humans and AI? Better together

As AI capabilities take over repetitive, more routine functions, human workers will be needed more than ever to not only manage and supervise technology but engage in higher cognitive functions, such as creativity, innovation, persuasion, and decision making.

Mercer describes this new, blended workforce as one that is “human-led and technology-enabled.” This idea of augmentation means that humans still matter in the workforce, despite fears to the contrary. While experts predict that 52% of existing human tasks will be performed by robots by 2025, technology will create 133 million new jobs by 2022.

According to Mark Sears, founder and CEO of CloudFactory, as quoted in Robotics Business Review, “People will always play important roles in the workplace of the future. Even in highly automated environments, people must continue to develop, train, and iterate machine learning models to facilitate robotic automation. The feedback provided by the humans in the loop—even in highly automated environments—is critical to the success of automation.”

Building an effective augmented workforce starts with learning

A high-functioning augmented workforce depends on effectively managing its two core elements: technology and people. But while the AI piece of augmentation depends on choosing the right technology, managing the human element isn’t as straightforward. The skills shortage means that organizations can’t simply hire their way to an agile, upskilled workforce. Instead, organizations must begin focusing on creating the workforce they need by engaging and retaining existing employees.

A successful augmented workforce requires highly trained, agile, continuously upskilled employees—employees who are both eager to learn and simultaneously given the opportunity to learn by their employers. HR has long known that learning is key to both employee engagement and retention.

In the age of digital transformation, learning has an additional, equally significant benefit: organizations that provide ongoing learning are anticipated to outperform those who don’t. Putting learning at the center of the organization enables the creation of a workforce that is always ready to work with and adapt to changing technology and the market, thus making the organization far more flexible and agile in an era of constant disruption.

The six keys to getting started

Building an augmented workforce begins with building a learning organization, one that enables employees to learn continuously, quickly upskill as needed, and drive their own careers paths.

For organizations accustomed to providing an employee stipend for a college course or offering one-day workshops, this new approach– learning at the center of the organization rather than at its periphery—can seem overwhelming. But while becoming a learning organization is crucial to remaining successful amid constant change, the transformation doesn’t have to happen overnight.

Organizations can begin with six manageable steps, efforts that help prioritize development and nurture learning for the long term:

1. Establish a learning culture. Start Thinking of learning as something that happens every day, within every job, for every employee, from hire to retire. Enable employees to try new skills and fail. In the workplace of the future, failure is an opportunity to move forward.

2. Nurture curiosity by offering learning opportunities to everyone. Research shows that every generation (not just Millennials) wants to learn. Offer learning opportunities to those in entry-level and leadership roles–and every role in between.

3. Enable learners to drive their own learning. Let employees follow their curiosity. Offer a wide range of accessible, ongoing, and free opportunities.

4. Develop learning agility. Learning agility isn’t an innate skill. To drive efficient upskilling, vanguard organizations are teaching employees how to learn effectively.

5. Create career paths that are not so much traditional as transformational. See beyond traditional hierarchies and realize that the careers of the future will be very different than those of today. By offering a wide variety of developmental experiences, organizations can inspire employees to define their own career paths (and simultaneously encourage them to stay).

6. Offer empowerment, not entanglement. Offering employees ongoing learning doesn’t guarantee they’ll stay. However, by offering learning portability, organizations create an environment of trust and employee empowerment. And as more organizations do the same, employees will be free to find the right work “home” and thus be more productive and engaged.

The convergence of learning and work in the digital era.

Who wins in the digital era? Organizations that use both people and technology resources efficiently—the augmented workforce. While the use of AI for routine tasks is on the rise, the abilities unique to human beings—the cognitive, the creative, the empathetic—are still crucial to an organization’s long-term success. How then can organizations help the human workforce make this shift, from manual tasks to cognitive and creative roles? By making ongoing learning central to work and placing development opportunities at the core of every role. Practically, this means creating an employee-centric learning environment, such as a multi-platform, agile methodology where the learner’s experience is the focus.

Supervised Stack Ensemble: Speeding Customer Service

Posted on 16/02/202015/06/2021 by Tushar

In the age of social media, companies are conscious about the reviews that are posted online. Any act of dissatisfaction can be meted out by way of tart sentiments on these platforms. And so enterprises strive hard to give 100% positive experience, by doing all that they can to address customer grievances and queries. But like they say, there are slips between the cup and the lip – not all grievances can be handled amicably.

Let’s take the specific case of call centers here. Their Service Level Agreement mentions terms like number of calls answered at a certain time of the day, percentage of calls answered within a specific waiting time, etc. Ensuring customer satisfaction and retention requires a far deeper, more holistic view of interaction between customer care representative (agent) and caller. There are other KPIs such as what causes a customer to be dissatisfied and number of escalations. But these seldom find a place in the SLA.

In this article, we will talk about identifying drivers of (dis)satisfaction and come up with ways to improve it. In the course, we will touch up on the solution design that can scale and institutionalize real-time decision making.

Introduction

We’ve all done it, dialing the call center for any issue encountered. We are surely an expressive bunch when it comes down to rattling our emotions and spitting out our dissatisfaction. And if that is not enough, we threaten to let our dissatisfaction be known to the rest of the world – through social media, not to mention #CustomerExperience.

While standard surveys exist to capture the sentiments of customers, the percentage of people filling these surveys is very low. This compounds the problem of effectively addressing customer needs.

Automating the task of predicting customer satisfaction requires a balanced mixture of text mining, audio mining, and machine learning. The resulting solution needs to:

Scale and be deployable
Identify the drivers of dissatisfaction
Generate actionable insights and generalize well to the population

Modeling Pipeline

Modeling pipeline includes all the components (data ingestors, model builder, model scorer) that are involved in model building and prediction. It is mandatory for the modeling pipeline to seamlessly integrate all the components for it to be scalable and deployable – production worthy. These components vary depending on the problem, available architecture, tools used, scale of the solution and turnaround time. The following pipeline was built in Google cloud to solve the problem of dissatisfaction in call centers.

Modeling (actual work – driver identification)

In the above problem, the satisfaction survey showed good internal consistency. Calls, emails and chats had sufficient discriminatory power to model customer satisfaction. Exploration of the data showed that the patterns were non-linear. However, like other psychometric models, the satisfaction model was plagued by three major issues which threatened its external consistency: shortage of data, variance and instability. These problems were addressed in the following manner:

First, the issue of data shortage was solved using resampling (bootstrapping). Second, the challenge of model instability was resolved using k-fold cross validation for tuning hyperparameters of different models. This was followed by model averaging. Finally, the issue of model variance was solved using stack ensemble approach on bootstrap samples. Several classification algorithms were used to build the first layer of the stack. Logistic regression was used to predict the outcome by combining the results from the first layer. The accuracy thus obtained was superior to that of any individual model in the first layer of the stack.

Driver Analysis

Only two types of classification models are directly interpretable: logistic regression and decision tree. Interpretation of other Machine Learning techniques such as regularized regression and regression splines require knowledge of calculus, geometry and optimization. Machine Learning models such as support vector machine and neural networks are considered black box techniques because of the high dimensionality, which is difficult for the human brain to comprehend.

Standard measures of variable importance exist for commonly used black box techniques such as SVM and neural networks. Simple weighted average method is used to calculate the importance of variables in the stack ensemble, with the weights being determined by the logistic layer. However, it is important to note that the final importance is not a measure of linear dependence of satisfaction on the independent variables. The importance metrics need to be combined with business intuition and actionability to provide recommendations for improving customer satisfaction.

Consumption

A call center manager would like to track customer satisfaction level along with several KPIs that are critical to operation. Information related to utilization of customer care representatives is provided to the manager in real-time. Model prediction is run in semi-real-time to reduce the required computational power. The manager is provided with options to deep dive into historical data based on variables that are drivers of dissatisfaction. For example, calls can be redirected to customer care representatives by existing ERP systems based on their history and subject matter expertise. This reduces the number of escalations and enables near real-time actionability without significantly affecting other KPIs.

The problem of customer dissatisfaction in call centers can be solved using audio mining, text mining and machine learning. Intelligent systems greatly reduce the stress on customer care representatives by automating majority of the processes. These cloud-based systems can be seamlessly integrated with existing ERP systems to provide highly actionable insights about dissatisfaction without significantly affecting other critical KPIs that are related to call center operations.

Industry 4.0 / Digital Transformation

Posted on 15/02/202015/06/2021 by Tushar

Digital Transformation / Industry 4.0 is on everyone’s mind. Investors are happy to hear from organizations that they are embarking upon a complete Digital Transformation / industry 4.0 journey. Investors love it, leaders advocate for it, directors have to make it a reality, managers have to design for it, but few understand what all it means in the grand scheme.

Hopefully, we can simplify this world for you.

What is Digital Transformation? Let’s keep it simple.

The simplest way to describe Digital Transformation is “Using Digital technology, innovation and intelligence to find better ways to do various things that organizations do today. It’s not about creating something new, but more about improving effectiveness and efficiency of existing processes for better business outcomes.”

Digital Transformation started as Industry 4.0 in some places. However, the idea remains the same. While Industry 4.0 started with the intention of transforming the manufacturing processes using Digital Technology, the principles of Digital Transformation now apply to all functions across the organization.

How does this theory apply in practice? Let’s study an example:

Step 1 – Current State

Map out the current process to uncover gaps that can be filled with better technology or intelligence.

Consider a global paper products manufacturing company. The manufacturing industry team is constantly trying to find opportunities to improve efficiency and productivity and reduce costs.

Energy consumption is a big area of focus for the manufacturing team. Currently, manufacturing industry reports and energy dashboards are used to track the consumption of energy across a few important machine parts.
Operators use these dashboards to identify sections of machines that are in green/red (good/bad) zones in terms of energy consumption and adjust the settings to optimize energy consumption.
These dashboards only track a limited set of machine parts that influence energy consumption.

Step 2 – Future State

Outline what the future should look like, after Digital Transformation.

Energy consumption of machines at the mill (specific reference to Tissue Machines) can be reduced by finding the key driving factors of energy consumption, determining their optimal settings while factoring in for the production constraints in terms of time, quantity and quality.

The following challenges will have to be addressed to get to the future state

There are a few hundred variables in a tissue machine that determine the energy consumption. These machine variables have to be studied comprehensively to identify the key influential factors for energy consumption. Relationships between these variables also need to be considered.
A detailed and statistically robust mechanism is created to generate insights/correlations across all relevant machine variables, to take proactive steps to minimize energy consumption.
Study the process characteristics that influence energy consumption and optimize them. E.g. machine speed, maintenance schedule, aging of parts.

Step 3 – Establish how technology, data and analytics can bridge this gap.

The best way Digital Transformation approach for this example would be:

Select a machine, in a market, which can be managed and monitored easily. Maturity in terms of capturing data, and the groundwork that has already been achieved for manufacturing systems and lean energy dashboards provides an immediate feasibility in terms of execution and adoption.
Build a Driver Model to understand key influential variables and determine the energy consumption profile.
1. Identify Key Variables –
  1. There are ~ 600 machine parts that drive the consumption of energy of a tissue machine. First, shortlist the top contenders and eliminate the non-influencer variables, using inputs from technical teams and plant operators.
  2. Identify primary drivers among the selected machine variables using variable reduction techniques of Machine Learning.
2. Driver Model –
  1. Building multivariate regression models to understand the impact of top drivers of energy consumption using techniques like linear regression, RIDGE/LASSO regression, Elastic Nets.
3. Optimize the engine to lower energy consumption.
  1. Optimize energy consumption by identifying the right combination of drivers under the given production constraints – time, quantity and quality.
  2. Create a mechanism to provide guidance during the actual production hours (In-line monitoring).
    1. Track energy consumption of the machine parts and their active energy consumption states. Identify deviation from the standards.
    2. In case of deviation, provide guidance to machine operators to bring the energy consumption to within defined limits.
4. Adoption
  1. Real-time dashboards, refreshed weekly, provide charts on energy consumption, recommendations, and improvements achieved through proactive measures.
  2. Post-live support to operations teams to enable adoption.
5. Scaling
  Determine phased roll-out to other machines using
  1. Strategic initiatives.
  2. Machines or mills which utilize higher amounts of energy to target higher ROI.
  3. Similarity in process and parts characteristics of tissue machines.
  4. Data availability and Quality.
  5. Readiness and groundwork for adoption by plant operators and energy management teams.

4 key stages in Digital Transformation

How should you, as a leader in an organization, look at Digital Transformation? Organizations should consider the 4 key stages of Digital Transformation, in order to create a sustainable impact on their organization. To make Digital Transformation a reality, all these steps cannot work independently. The philosophies of Design Thinking are embedded in the framework’s interconnected elements.

DEVELOPMENT PHASE:

Focus is on identifying the key areas and prioritizing the Digital Transformation efforts

Stage 1 – Discovery

Identify the key areas of opportunity or risk and related key stakeholders. Detail out the gaps in process, data, insights or technology, fixing which would help capture opportunities or mitigate risks.

Stage 2 – Design

Rapid iterations on design and implementation of prototypes helps reach optimal solutions faster. Build out Proofs of Concepts (PoC) to establish the theoretical validity of the approach. Validate the practical validity of the approach through a Proof of Value (PoV).

IMPLEMENTATION PHASE:

Implementation needs to account for limitations arising from human behavior and scale of the operations.

Stage 3 – Adoption

Building solutions that keep the user at the center of the design, is key to adoption. This means that users must be included in the design and feedback early on. In addition, there should be support for users post design, in the form of FAQs, training videos, chatbots etc.

Stage 4 – Scalability

If we can’t solve a problem at scale, then the solution does not solve organizational problems. The issues that we anticipate at scale, should be accounted into the design in the Development phase. This means considering the technology used, the infrastructure required, process automation possible / required and how to manage future developments.

Like Design Thinking would dictate, the Development phase of the Digital Transformation processes have to always consider the Implementation aspects.

Digital Transformation is no longer just optional.

Every organization is transforming the way they do business. Numerous organizations like BASF, Mondelez, KLM airlines, Aptar group, PepsiCo etc. are already making massive strides in this area.

If you want to zip past your competition, or even stay competitive, it’s about time you started thinking about how to transform the way to do business. After all, there’s no growth in comfort.

AI Applications in Documents

Posted on 13/02/202015/06/2021 by Tushar

“We are drowning in information, but starved for knowledge”

This is a famous quote by John Naisbitt which shows the key difference between information and knowledge. Advancement in data engineering techniques and cloud computing have made it easy to generate data from multiple sources but making sense of this data and getting insights is still a huge challenge. The data volumes have now increased exponentially and along with the traditional structured data, data can now reside in different formats like unstructured social media text, log files, audio/video files, streaming sensor data etc.

Applying manual methods to process this diverse data is not only time consuming and expensive but is also prone to errors. Hence the need of the hour is to use Artificial Intelligence (AI) based automated solutions that can deliver reliable insights and also give a competitive advantage to customers. Here are few examples of how customers across industries can benefit from AI driven solutions.

Microsoft Azure based AI solution

In 2017, more than 34,000 documents related to John F Kennedy’s assassination were released. The data volume was huge, and data existed in different formats like reference documents, scanned PDF files, hand written notes and images. It would take researchers months to read through this information and hence manually reviewing this data was not the most optimal solution. Microsoft Azure team applied AI based Cognitive Search solution to extract data from these diverse sources and gained insights. Technical architecture for this use case was built using Azure Cognitive Services components like Computer Vision, Face Detection, OCR, Handwriting Recognition, Search and core Azure components like Blob Storage, Azure ML, Azure Functions and Cosmos Database. This solution also annotated text using custom CIA Cryptonyms.

Hospitals usually deal with a lot of patient data which could reside in electronic medical records (EMR), handwritten prescriptions, diagnostic reports and scanned images. AI based Azure Cognitive Search could be an ideal solution to efficiently manage patient’s medical records and create personalized treatment plan. Many downstream use cases like Digital Consultations, Virtual Nurses and Precision Medication can be built once the patient data is optimally stored.

Google Cloud Platform (GCP) based AI solution

GCP introduced Document Understanding AI (beta) in Cloud Next 19. This is a serverless platform that can automate document processing workflows by processing data stored in different formats and building relationships between them. This solution uses GCP’s vision API, AutoML, machine learning based classification, OCR to process image data and custom knowledge graph to store and visualize the results. Customers can easily integrate this solution with downstream applications like chatbot, voice assistants and traditional BI to better understand their data.

Customers who deal with Contract Management data like Mortgages are usually faced with a lot of manual tasks to ensure that the contracts are complete and accurate. This could mean processing contracts in different formats/languages, reviewing the supporting documents, ensuring that the details are accurate and complies with regulatory standards across documents. By using Document Understanding AI and integrating it with a well-designed RPA framework, customers will be able to efficiently process Mortgage applications, Contracts, Invoices/Receipts, Claims, Underwriting and Credit Reports.

Use cases from other industries

Document Management AI solution can also be applied to diverse use cases from other industries like processing claims related to damages to shipped products by e-commerce companies, handling know your customer (KYC) process in the banking industry, invoice data processing by Finance teams, fraud detection during document processing etc.

As more and more companies embrace the digitization wave, they will be faced with different variations of data/document management challenges. Based on the current trend, number of use cases are only going to increase and an AI driven solution is probably the most efficient way to solve this problem as it can reduce manual work, save cost and deliver reliable insights. This will ensure that companies can spend more time on building their business and less time on manually processing documents and data preparation.

The New Customer Satisfaction Era

Posted on 12/02/202015/06/2021 by Tushar

Let us start with an oft repeated question,” What do you know about your customer’s preferences”?

The answer could be any of the standard responses which talk about their tastes in your merchandise based on past transactional records. It could be also one of the slightly more personalised answers which talk about the customer’s likes and dislikes basis whatever they have filled in their surveys and feedback forms. Does this tell you all you need to know about your customers? Does this help you make the customer experience of that customer something which he/she will remember? Something that gets ingrained into the sub-conscious decision-making component of their minds. That is the holy grail which most CX organisations are after.

Where does data come into the picture?

With 91 properties around the world, in a wide variety of locations, the Ritz-Carlton has a particularly strong need to ensure their best practices are spread companywide. If, for example, an employee from their Stockholm hotel comes up with a more effective way to manage front desk staffing for busiest check-in times, it only makes sense to consider that approach when the same challenge comes up at a hotel in Tokyo. This is where the hotel group’s innovation database comes in. The Ritz-Carlton’s employees must use this system to share tried and tested ideas that improve customer experience. Properties can submit ideas and implement suggestions from other locations facing similar challenges. The database currently includes over 1,000 innovative practices, each of them tested on a property before contributing to the system. Ritz-Carlton is widely considered to be a global leader in CX practises and companies like Apple have designed their CX philosophy after studying how Ritz Carlton operate.

What does this tell you- Use your Data wisely!

The next question that may pop up is, “but there is so much data. It is like noise”. This is where programmatic approaches to analysing data pop up. Analytics and data sciences firms across the globe have refined the art of deriving insights out of seemingly unconnected data to a nicety. What you can get out of this is in addition to analysing customer footprint in your business place, you get to analyse the customer footprint across various other channels and social media platforms.

This aims to profile the customers who are most susceptible to local deals/rewards/coupons basis their buying patterns.

How is this done? The answer is rather simple. Customer segmentation algorithms (both supervised and unsupervised) enable you to piece together random pieces of information about the customer and analyse the effect they have on a target event. You will be surprised at the insights that get thrown out of this exercise. Obviously caution needs to be exercised to ensure that the marketeer doesn’t get carried away by random events which are purely driven by chance.

Okay- so I have made some sense out of my data. But this is a rather cumbersome process which does not make any difference to the way I deal with my customer on a day-to-day basis.

“How do I get this information on a real-time basis so that I can actually make some decisions to improve my customer’s experience as and when it is applicable?”

This takes into the newest and most relevant trend into making data sciences a mainstream part of decision making. How do we integrate this insight deriving platform into the client’s CRM system so that the client can make efficient decisions on a real time basis?

In Anteelo, for one of our leading technology clients, we have built an AI-based orchestration platform which derives the actionable insights from past customer data and integrates this into the customer’s CRM system so this becomes readily available to all marketeers as and when they attempt to send out a communication to their customers.

What does this entail? This entails using the right technology stack to build a system which can delver insights from the data science modules at scale. I prefer calling it out as a synergy of both data sciences and software development. Every decision that a marketeer is trying to make must be processed through a system which will invoke the DS algorithms in-built on a real time through the relevant cloud computing platforms. Insights will be delivered immediately, and suitable recommendations will also be made on a real-time basis.

This is the final step in ensuring that personalised recommendations being made to every customer are truly personalised. We in Anteelo call it “The Last Mile adoption”. This development is still in its nascent phase. However, companies would be wise to integrate this methodology as a part of their data science integrated decision making since it is very unlikely that they will hit the holy grail of customer satisfaction without delivering real-time personalised recommendations.

What Does Product Management Look Like in Data Science?

Posted on 11/02/202015/06/2021 by Tushar

The topic related to ‘Product Management’ has received several laurels in recent years. Several rounds of discussions have happened to create an analogy out of client’s stand point. As I heard more of these conversations, there was an uncomfortable ambiguity stemming from disbelief – is this another fad or is there meaning to it? Well, the initial rumblings were from the cool kids in the bay. But, why did grounded Midwest and shoot-from-the-hip south latch on? Must be something deeper, right?!

Product management has been around forever in the software, e-commerce world. But, today, mainstream IT and AI teams in fortune 500 companies are thinking of a product paradigm shift. Leading consulting firms are also developing products or beefing up their technology as an eventuality.

But, the question that begs attention here is – why products? What happened to software as a service, platform as a service, ML as a service? Do we need another paradigm shift? Or as the saying goes – Old wine in a new bottle?

IT teams are today being led by progressive Chief Digital Officers, Chief Data officers. Conventionally, CIOs have been leveraging their value by app dev teams, BI teams, infrastructure teams et al. While this may have become a table stake, it has been around for a while already. The question is – ‘How to deliver incremental value to business?’

So, what has changed?

Demand:

IT is today called upon to be a true business partner. And, given the rate at which business is facing change, the time to deliver value is compressed.

Glocal innovation:

For a fortune 500 firm operating globally, innovation is striking at its core from multiple directions. While the USA is still the biggest revenue (EBITDA generation engine), problem and solution innovation is happening in other markets faster than the USA. For starters, they have less legacy to deal with. The local markets are facing competition from nimbler players. VC money is flowing into firms in China, Israel, Korea, India which are encountering newer problems in e-commerce, voice commerce sectors. Other traditional revenue generating markets, individually facing slower growth, find it difficult to make a business case to invest in solutions led by such innovations.

Problem repeatability:

This is going to sound rhetorical. But, I must state it because it is relevant. Business problems in today’s enterprise are constantly changing. Few of them get recreated, and hence are not available in large volumes. Few others are becoming common across markets and thus moving into a constant state of being a tightly defined problem that can be applied globally. Repeatable.

A good indicator to this is AWS recent product launches – out of the box image, text, voice, reinforcement learning, forecasting. Common problems which are finding repeatable solutions.

The AI candy shop:

Today, nobody wants to use process automation tools that are not embedded in intelligence. Passé, inefficient. Wallstreet, investors and boards are lapping up the buzzwords – cognitive, AI, embedded ML.

Cloud enabling global scalability:

Cloud platforms such as Azure, AWS have ensured that once you have these AI capabilities developed, they can be deployed globally. The global-local adaptation is a key design criterion in this context.

Glocal solution adaptation…er,… maybe Glocal problem adaptation:

Each market has its secret sauce in terms of the market structure, drivers and customer nuances. Thus, before adapting a solution from one market to the other, it is essential to adapt the problem as well. For example, it is an interesting pursuit to adapt the problem structure from the modern trade Australia market to half way across the world in Brazil.

And, then adapt the solution.

So, who’s game is it anyway?

Given the above guard rails, it is quite evident that the business case should be developed by a country specific P&L or ROI measure. It must be a global mandate. IT is one of the few functions which is ideally poised to ride this wave. That, they own the data systems is coincidental. Or, well.. was that the whole plan! Go, Trojan..

Finally, after rambling about half the things in the world – we come to the initial topic of this article. Products. Why?

A product has a life – it evolves constantly. The focus is on continually making the best product for its end user, ever possible. It has a roadmap. In a world of multiple users, it needs a strong owner who plans and decides well. It has a clear value proposition in each update/release. It can be developed in a sprint like manner. It can be defined with a bounded scope and sold internally in enterprises, with greater ease. And, be defined, abstracted, customized for a global roll out.

Looks like a duck, walks like a duck, sounds like a… must be a duck. Yes, I guess it does look like a product.

But, how do we help organize people and teams to get the products rolled out?

While the below roles are common to a product-oriented firm, the thought process is different from conventional IT projects. Sharing of resources across projects being the biggest drawback. The smartest of each of the below folks will perhaps still fail, without an organizing framework. The roles to work in a closely integrated manner, dedicated to making a single product management successful.

Product Designer:

The role of a product designer is someone who can completely immerse himself in the shoes of the end user, without worrying about the AI or Tech related issues that may occur sometimes. Just solve the end user’s real problem and keep tracking the end user’s behaviour as the product usage evolves. In product management, there is a contradictory school of thought which mandates that the designer must appreciate “how” a product works. This, however, might dilute the designer’s objective of empathizing with the end user.

Product owner:

A functional expert of impact analysis who can connect the dots and identify the nuances of each problem. A great problem solver, with functional expertise, has the knack to see through the commonalities, and the uncommon aspects too. Prioritization between the must-haves, nice-to-haves and must-not-haves is a key skill required in the role.

Product BAs

Products are quite massive in terms of their scope today. Primarily, each product usually is broken down into sub products which are owned by individual product Bas.

The AI solution developer(s)

Usually, it is very difficult to get a product owner who really gets AI solution development. By and large, individual intelligence is anyways overrated. It is important to have a dedicated AI solutioning team which can translate the problem into a modular AI solution.

The AI deployment team

It is not enough to develop a modular AI solution. To be able to deploy it in globally scalable platforms requires seasoned IT big data engineering & testing capabilities. The plumbing and wiring required to take the AI concept to enterprise last mile reality is no mean task. It is a specialized function. Truly speaking, they give the product its real-life form.

Scrum & Program Managers

Last but not the least, you need the scrum team and program managers. Everyone benefits from their discipline and order amidst the chaos.

The growing popularity of low-code/no-code application development platforms

Posted on 09/02/202014/06/2021 by Tushar

“Software is eating the world.” That was the bold proclamation renowned innovator and venture capitalist Marc Andreessen expressed in an article he wrote for the Wall Street Journal. More and more businesses are being run by software, he argued, or are differentiating themselves and disrupting their competitors and industries using the same. Today, that article is regarded as one of the seminal works in shaping how people think about digital transformation—or using digital technologies to bring about great change in the way individuals and organizations think, operate, communicate or collaborate.

Often, at the heart of many digital transformation efforts is the desire to enable the organization to be more agile or responsive to change. This requires looking for ways to dramatically reduce the time needed to develop and deploy software, and simplify and optimize the processes around the maintenance of software so it can be deployed quickly and with greater efficiency.

Another key outcome that is part of many digital transformation efforts is enabling the organization to be more innovative — finding ways to transform how the organization operates and realize dramatic improvements in efficiency or effectiveness; or creating new value by either delivering new products and services or creating new business models.

For organizations using conventional approaches to developing software, this can be a tall order. Developing new applications can take too long or require very specialized and expensive skills that are in short supply or hard to retain. Maintaining existing programs can be daunting as well, as they struggle with increasing complexity and the weight of mounting technical debt.

Enter “low-code” or “no-code” application development platforms. This emerging category of software provides organizations with an easier to understand — often visual — declarative style of software deployment, augmented by a simpler maintenance and deployment model.

Essentially these tools allow developers, or even non-developers, to build applications quickly, easily, and rapidly on an on-going basis. Unlike Rapid Application Development (RAD) tools of the past, they are often offered-as-a-service and accessed via the cloud, with ready integrations to various data sources and other applications (often via RESTful APIs) available out of the box. They also come with integrated tools for application lifecycle management, such as versioning, testing, and deployment.

With these new platforms, organizations can realize three things:

1. Faster time to value

The more intuitive nature of these platforms allows organizations to quickly get started and create functional prototypes without having to code from scratch. Pre-built and reusable templates of common application patterns are often provided, allowing developers to create new applications in hours or days, rather than weeks or months. When coupled with agile development approaches, these platforms allow developers to move though the process of ideating , prototyping, testing, releasing and refining more quickly than they would otherwise do with conventional application development approaches.

2. Greater efficiency at scale

Low-code/no-code application development platforms allow developers to focus on building the unique or differentiating functionality of their applications and not worry about basic underlying services/functionality such as authentication, user management, data retrieval and manipulation, integration, reporting, device-specific optimization, and others.

These platforms also provide tools for developers to easily manage the user interface, data model, business rules and definitions, making on-going management easy and straightforward. So easy in fact that even less experienced developers can do it themselves, lessening the need for costly or hard-to-find expert developers. These tools also insulate the need for the developer and operations folks to keep updating the frameworks, infrastructure and other underlying technology behind the application, as the platform provider manages these themselves.

3. Innovative Thinking

Software development is a highly creative and iterative process. Using low-code or no-code development platforms, in combination with user-centric approaches such as design thinking, organizations can rapidly bring an idea to pilot in order to get early user feedback or market validation without spending too much time and effort (so-called “Minimum Viable product” as coined by Eric Ries in his book “The Lean Startup”).

Not only that, because these platforms make it easy to get started, even non-professional developers or “citizen developers,” who more likely than not have a deeper or more intimate understanding of the business and end user or customer needs, can develop the MVP themselves. This allows the organization to translate ideas to action much faster and innovate on a wider scale.

While offering a lot of benefits, low-code/no-code application development platforms are certainly not a wholesale replacement to conventional application development methods (at least not yet). There are still situations where full control of the technology stack can benefit the organization—especially if it’s the anchor or foundation of the business, the source of differentiation, or source of competitive advantage. However, in most cases organizations will benefit from having these types of platforms as part of their toolbox, especially as they embark on any digital transformation journey.