Customer support operations are crucial to business success, yet many face challenges with inefficient processes, disconnected workflows, and existing agent-routing systems that are inefficient when connecting users to agents with appropriate skills
The traditional chat system relies on “canned” buttons with the same predefined text that serve all user interfaces. Support solutions do not alter based on what a specific user situation or need could be, which leaves little room for innovation and creates a void for personalisation. The outcome is often longer resolution times, diminished productivity, and missed growth opportunities.
The most relevant question we must address in the realm of customer support is “How can we tailor our conventional systems to leverage the user information in a real-time manner to provide a stellar class support experience for our customers? Can real-time context be embedded into workflows and interactions for better outcomes?
According to a report from McKinsey, more than 71% of the customer base expect personalized interactions from their companies.
As a Lead Android Engineer with over a decade of experience working on consumer electronics and user experience enhancements products, I’ve tackled challenges head-on that push the boundaries to create a solution that caters to every customer’s need uniquely. While working on a business growth vision initiative, I developed a dynamic user interface chat component that analyses the user’s screen identifier and maps the component to relevant support topics via an intelligent support identification mechanism, yielding improved user engagement and a cost-savvy business outcome.
While at Navan (formerly TripActions), I co-designed a context-based customer interaction system that was later patented as a part of TripActions’ public patents. This innovation reduced handle time for agent user interactions by 40–50% and redefined enterprise workflows by dynamically tailoring user interfaces to individual needs in real-time.
Drawing from this experience, implementing this support system during the COVID-19 pandemic made me realize the value of this transformative application was for the unprecedented volume of support inflow we had to manage.
Why Context Matters in Contact Center Operations
When I first began working with customer support solution platforms, I noticed a recurring issue: Inefficient agent matching, inadequate problem identification and redirection to static articles that were not satisfying users’ specific concerns.
Traditional chat systems lacked the ability to adapt dynamically to users’ unique contexts, compounding the problem further.
Intelligent contextual chat systems bridge this gap by leveraging real-time data, predictive analytics, and adaptive user interfaces. These systems embed context into every interaction, enabling smoother workflows, faster problem resolution, and better decision-making.
At Navan, for instance, I implemented a Twilio SDK system to integrate real-time data and dynamically predict users’ needs. This innovation reduced the need for agent assistance and significantly improved customer satisfaction metrics.
Such approaches demonstrate how contextual systems can transform operations, drive efficiency, and deliver measurable business outcomes.
Technical Architecture of Intelligent Contextual Chat Systems
The development of intelligent contextual chat systems requires a strong technical infrastructure. Drawing from the patent, here is a breakdown of the core components and their functions:
1. System Components
A well-structured system relies on several core components to ensure seamless functionality. Below are the key elements that make up the system:
- User Interface Display Component (103):
Executes on a server-side computing device (106a). It scans URLs and screen IDs, retrieves relevant data, and dynamically generates UI elements. Additionally, it maintains mappings between screens and support topics while filtering content based on user context. - Routing Component (111):
Runs on a separate computing device (106b). It manages the association between UI elements and agent attributes, incorporates selection algorithms with fallback mechanisms, and facilitates network connections between users and agents. - Profile Database (120):
Stores critical user data, including account information, travel reservations, and corporate records. It is implemented as an ODBC-compliant database, supporting platforms like Oracle, Microsoft SQL, SQLite, PostgreSQL, MySQL, NoSQL, and custom databases.
These components work together to deliver an efficient, responsive system that enhances user experience and optimizes operational processes.
2. Technical Implementation Details
The system’s technical implementation ensures seamless data flow and accurate user interactions. Here’s how it works:
- URL/Screen Analysis:
The browser sends the current URL to the user interface display component. The system detects changes during on-screen events for single-page applications, while for mobile apps, it dynamically analyzes screen or view identifiers. - Support Topic Mapping:
The system uses structured data mappings (tables) to associate URLs with relevant support topics. To ensure accurate assistance, it maintains relationships between screen identifiers and corresponding topic lists. - User Profile Access:
The system performs database queries by using unique user identifiers from account logins. Some implementations also incorporate biometric data for added security and authentication. - Reservation Data Handling:
A dedicated computing device (106c) periodically checks for updates from external sources. Trip details and status updates are stored within user profiles to support real-time adjustments.
These technical processes work together to provide a smooth and intelligent user experience, adapting dynamically to different environments.
These components create a responsive, context-aware system that enhances user experiences while optimizing operational efficiency.
Practical Applications in Engineering
Contextual systems have a profound impact on engineering workflows. In my experience, they deliver:
Enhanced Developer Productivity
By integrating real-time data analytics and adaptive user interfaces, systems are developed to dynamically adjust to user needs, reducing the need for manual interventions and streamlining customer support workflows. This enables faster issue resolution, improves agent efficiency, and allows teams to focus on delivering relevant, context-driven solutions.
As the system dynamically adapts to users’ behavior and context (such as booking details or real-time data), it reduces friction and accelerates response times, enhancing overall support performance. This focus on dynamic adaptability not only improves user experience but also ensures scalability as the user base grows, enabling businesses to efficiently handle increasing demands without sacrificing quality.
Reducing Operational Costs with Contextual Automation
One of the key benefits of contextual systems is their ability to automate repetitive tasks and streamline operations, resulting in significant cost reductions. For example, AI-powered chatbots can handle a large volume of basic customer queries, reducing the need for human intervention and thereby lowering operational costs.
In engineering, this automation extends beyond just customer support—automating incident management and error resolution processes can minimize downtime and reduce the need for manual oversight. Contextual automation systems continuously monitor user activity, analyze real-time data, and adapt to provide proactive solutions to potential issues, all of which contribute to lowering costs across various departments.
Additionally, by integrating AI and machine learning with contextual systems, we can optimize resource allocation, anticipate future demand, and fine-tune systems for maximum efficiency, further reducing costs and improving business outcomes.
Improved Cross-Team Collaboration
Effective collaboration is at the heart of successful engineering workflows. I facilitated smoother communication between product, QA, design, backend, and customer support teams across organizations by embedding contextual systems.
For instance, at Navan, these systems ensured that all stakeholders had access to real-time contextual information, enabling efficient feature development and timely releases. This cross-functional alignment was pivotal in driving innovation and achieving measurable outcomes, such as a 40x growth in the user base and increased customer retention.
Contextual systems are not just tools for improving engineering operations; they are catalysts for driving business success. From reducing inefficiencies to fostering collaboration and innovation, their impact extends beyond technical workflows, creating value at every level of an organization.
While contextual systems offer immense potential, they present challenges such as real-time data integration, Accurate context Identification, and System performance. Here’s how I’ve addressed them:
- I implemented a Kafka-based event streaming platform to provide support for processing and distributing real-time events, ensuring contextual chat components are fed with the latest events for any change in flight status, user profiles, or reservations status.
- We coded a multi-signal context resolution engine which included user interaction patterns, screen path traverse history, and processed through a weighted decision tree to determine the user’s support intent and thus distinguishing between multiple users viewing the same screen vs the support journey of an individual contacting support.
- To improve network optimization for contextual data transmission, I implemented a Protobuf serialization that dramatically reduces payload size by encoding travel context data in a compact binary format rather than verbose JSON text, enabling faster network transmission and processing on mobile devices.
The Future of Contextual Chat Systems
Looking ahead, I envision contextual systems integrating with emerging technologies like AR/VR, which could revolutionize training and simulation environments for engineering teams.
Imagine immersive, context-driven virtual scenarios that allow engineers to troubleshoot complex issues or test solutions in real-time, fostering innovation and reducing costly trial-and-error processes.
Beyond AR/VR, these systems will evolve into proactive assistants, leveraging predictive analytics to anticipate user needs and address potential issues before they arise.
For instance, such systems could preemptively identify workflow bottlenecks or provide tailored suggestions to optimize resource allocation, further enhancing operational efficiency.
Simultaneously, contextual systems have the potential to play a pivotal role in supporting sustainability initiatives. My work on Navan’s Carbon Impact Tracker demonstrated how real-time personalized insights could empower businesses and users to make environmentally conscious decisions, such as reducing the carbon footprint of business travel.
This highlights the broader applications of these systems in driving sustainability while aligning with organizational goals.
Leadership Lessons
My career taught me that collaboration and scalability are key to building impactful systems. Whether leading projects at JPMorgan Chase or overseeing Mason OS, I’ve prioritized cross-functional alignment and user-centric design.
By blending technical innovation with business acumen, I’ve created systems that empower organizations to thrive in a digital-first world.
Final Thoughts: AI-Powered Contextual Engineering
Intelligent contextual AI chat systems are no longer just a support tool—they are driving engineering innovation, operational efficiency, and business growth.
By enhancing developer productivity, optimizing incident management, and enabling scalable engineering solutions, these AI-driven systems are paving the way for the future of intelligent enterprise automation.
As AI evolves, organisations must embrace context-driven automation to stay ahead in an increasingly intelligent and connected world.
Reference
Patent Google. (2021). Methods and Systems for Dynamically Generating Contextual User Interface Elements. https://patents.google.com/patent/EP4179426A1/en?oq=0EP4179426A1
SpringerLink. (2020). Building Contextual User Interface Systems. Retrieved from https://link.springer.com/article/10.1007/s12525-019-00359-6
Google Books. (2017). Contextual Chat Systems: Drive Efficiency and Innovation. https://books.google.com.ph/books?hl=en&lr=&id=rdMVBQAAQBAJ&oi=fnd&pg=PP1&dq=Contextual+Chat+Systems+Drive+Efficiency+and+Innovation&ots=wDMCTT0TE_&sig=4FL7tPEfVZ6ZA_wlZYtidcCa_0Q&redir_esc=y#v=onepage&q&f=false
IEEE Xplore. (2013). Dynamic Contextual Systems for Contact Centers. Retrieved from https://ieeexplore.ieee.org/abstract/document/6512846
ResearchGate. (2023). Building Intelligent Chatbots: Tools, Technologies, and Approaches. Retrieved from https://www.researchgate.net/publication/371755945_Building_Intelligent_Chatbots_Tools_Technologies_and_Approaches
IEEE Xplore. (2016). Innovative Methods in Dynamic Chat Systems. Retrieved from https://ieeexplore.ieee.org/abstract/document/7451279
arXiv. (2020). Contextual Chat Systems in IoT Applications. Retrieved from https://arxiv.org/abs/2007.03051
(Top, Featured Image via: Shutterstock)
