Global customer experience and technology consultancy services company CSS Corp helps brands reimagine customer engagements. It helps clients manage over 1.2 billion interactions and, in the process, provides technologies, tools and solutions to enhance customer engagements. Its area of focus includes shifting from reactive to pre-emptive modes and helping companies adopt the right strategies to bring in efficiencies and delivery models optimisation.
Recently, Analytics India Magazine got into a conversation with Dr Kiran Marri, Chief Scientist and Vice President of Digital Engineering at CSS Corp, to understand how CSS Corp is leveraging artificial intelligence and machine learning.
Kiran is an industry consultant, educator, and researcher with over 18 years of experience in software and systems. An alumnus of IIT Madras (BE, MS and PhD), Kiran’s current role at CSS Corp includes developing innovative solutions, creating sustainable business models for growth, and launching new services in the digital, engineering and testing practices. He is currently researching to apply ML techniques and optimisation algorithms to solve complex engineering and software problems, build solutions for early validation strategies, and develop UX tools for global clients.
Edited excerpts from the interaction:
AIM: What are CSS Corp’s core tech products?
Dr Kiran: We have reimagined customer engagements and have enabled over a billion customer service and support transactions in over 25 languages across 19 global locations. We help clients succeed through intelligent automation-led outcomes and customer-centric solutions & services with the help of two AI-driven platforms:
- CSS EDISON: AI-led cloud-based CX transformation platform.
- CSS CONTELLI: AIOps platform; it takes an outcome-based approach and leverages cognitive RPA, AI, automation, augmented reality, and advanced analytics for IT support and services.
- COEUS: provides real-time insights into agent performance and helps improve overall efficiency and productivity. Last year, COEUS analysed over 10 million transactions, resulting in a 50 per cent improvement in service levels across accounts and increasing employee productivity levels by 25 per cent.
These platforms cut across the front and back offices of any enterprise to drive meaningful business outcomes.
AIM: What tech tools do you use?
Dr Kiran: Our tools vary depending on the services. We use a broad mix of proprietary home-grown tools as well as industry-standard tech tools.
AIM: How are you leveraging AI and ML technologies?
Dr Kiran: CSS Corp proposes a set of solution elements that are flexible and extensible – supported by optimum service delivery model and usage of next-generation tools, methodologies, and framework. Our framework for AI/ML projects across domains is based on five fundamental tenets:
- Design right talent sourcing strategy for the program
- Define a CoE model that will provide utmost operational flexibility and tight alignment to business goals while achieving the benefits of consolidation and standardisation
- Enable Service Transformation that consists of tools, accelerators, new practices, and solution components, and part of ‘managed services’ delivery model, driven by SLA framework
- Implement an execution framework based on CSS Corp offerings that will enable rapid and de-risked transition, process standardisation and fosters innovation
- Implement Program Management and Governance guided by KPIs, metrics, and stakeholder specific dashboard
AIM: Provide us some real-life use cases.
Dr Kiran: Here are some use cases:
- One of our US-based OEMs faced difficulty detecting the quality of images (video streaming) in UCAAS products. We improved their quality benchmarking using our AI/ML and computer vision solutions.
- A Europe-based CPF company had challenges in detecting faulty image uploads and automating the process flow. With our cutting-edge solutions that leverage AI/ML, computer vision, and customer algorithms, we enhanced their user experience and quality.
| Problem Statement | Domain | Client | Drivers |
| Automated detection of POI on the road help in ticket creation | GIS | Map provider | Improve user experience.Build an automated, cost-effective solution. |
| Map structure comparison – details, layout | GIS | Map provider | Benchmarking and Competition analysis.Improve user experience. |
| Automated extraction of data from images and create base for POI | GIS | Map provider | Reduce cost.Improve quality. |
| Automated extraction of data from scanned documents | Aggregator | Large taxi aggregator | Maximize throughput.Improve quality. |
| Detect faulty image uploads and automate the process flow | Retail | European based FMCG | Improve user experience and quality. |
| Automated correction and removal of background in large image dataset | Multiple | Banking client | Standardization & Automation |
| Detect quality of images (video streaming) in UCAAS products | OEM | OEM client based in the US | Quality benchmarking |
AIM: How large is your analytics team? What are the non-negotiable skills in data scientists?
Dr Kiran: We have a team of 130+ data scientists, data analysts and engineers. We hire data scientists for predominantly two purposes — client projects and research.
While hiring a data scientist for client projects, the prerequisites are a good understanding and knowledge of statistics, programming, ML, principles of computer vision, signal processing, and imaging techniques. The client projects can vary based on the type of data/combination (text, images, videos, or audio). So, in-depth knowledge or expertise based on the project scope is an added advantage.
For lab research, we look for data scientists who love problem-solving and applying scientific methodology to solve business problems. This role requires understanding the pain points, scoping them, identifying the constraints, evaluating the pros and cons, and ideating the most practical, scalable, and cost-effective solution.
AIM: Which sectors are you catering to? Explain the workflow using a case study.
Dr Kiran: We have a strong presence in retail, communications, media and technology, consumer electronics, education, healthcare, manufacturing, and geospatial (telecom, utilities and navigation).
Case Study:
With our AI/ML expertise and capabilities, we automated image classification for an e-commerce client.
Problem Statement:
- Identifying image and classifying takes a long time and is highly expensive
- 20 per cent of images were not in-store images, and turn-around time over ten days
Scope of work:
- Automate the end-to-end processing of image identification and classification
- Handle multiple image formats
- Automatic rejection of non-store images and low-quality images
Approach:
- Image identification through image histogram, structural similarity, and pattern recognition
- Machine learning models, computer vision principles, image processing, image enhancement algorithms
Features and benefits:
- Multi-model image identification and classification
- Over 95 per cent accuracy in identifying non-store images
- Productivity improvement by more than 70 per cent turn-around time for failed images reduced by up to 90 per cent
- Manual intervention is required only on a small portion of images with low confidence classification
AIM: What are the tech trends in the digital customer service space?
Dr Kiran: The pandemic-induced disruption ignited radical shifts in customer service and contact centre operations. Here are a few trends that will drive digital customer service:
- Cloud is a given, not an option anymore: Cloud-based solutions deliver innovative digital contact flows rapidly and match changing consumer preferences. It also increases flexibility and scale of operations as needed and can be leveraged as a pay-as-you-go model.
- Digital messaging is the new conduit for the entire support experience: Twitter, FB Messenger, Instagram, SMS, WhatsApp, Apple Business Chat — all offer opportunities to enhance customer experience and interactions. Customer experience leaders leveraging asynchronous two-way mobile messaging and notifications can help customers proactively.
- Channel-less environments: Frictionless crossovers: Before the disruption, customer service tended to be more reactive, burdening customers to find the best channel for getting a solution themselves. But, this has changed now. Customers do not care about a channel but the experience. Hence, it is imperative to create an infrastructure that blurs the boundaries between the distinct channels and creates an uninhibited experience for customers.
- Data and insights drive proactive, personalised experiences: Customers expect service interactions tailored to their personas, transaction and interaction histories, and current states. Companies need to leverage interaction analytics to identify sales opportunities, proactively trigger customer retention workflows, and enable agents with real-time insights.
- Strong inclination towards AI will lead the way: AI cannot entirely replace 1-on-1 human interaction in the customer service industry despite many years of buzz. It’s worth mentioning that it takes two to tango to offer exceptional customer service — human agents and AI. AI is not here to replace humans. Instead, it helps them work more efficiently. For example, AI-powered technology can reduce handle time by auto-populating call notes or automatically logging agents into or out of applications to save time. AI also enables better connectivity for customer service teams and allows agents to receive consistent communications and information to excel in serving customers.
AIM: What are CSS Corp’s plans for the future?
Dr Kiran: We are investing heavily in new-age technologies. In the coming quarters, we plan to expand our GIS and Tech/Customer Support. In the former, we have already started leveraging AI/ML to ease the operations. In the latter, we are increasing process automation and AR/VR based solutions for support engineers.
