Use Cases for Ved: GenAI-powered ChatBot for Site Reliability Engineers and Business Operations
- Aug 7, 2024
- Blogs
- 5 min read
The introduction of Ved, our GenAI-powered chatbot, aims to transform observability for Site Reliability Engineers (SREs) and Business Operations teams. By leveraging advanced machine learning and natural language processing, Ved simplifies complex tasks and enhances real-time decision-making.
Building on the technical foundation detailed in our previous blog on Ved’s GenAI architecture, this post explores practical applications and key differentiators that set Ved apart in the enterprise space.
Despite the nascent state of this technology, we believe that the intersection of domain-centric knowledge and data, combined with powerful LLM models, will drastically simplify the last-mile usage of observability platforms.
Ved’s ability to interpret and respond to natural language queries makes it accessible to a broader range of personas, from SREs and DevOps to business operations teams. This reduces the need for extensive platform training and constant monitoring of dashboards and alerts.
Given that Ved leverages the RAG (Retrieval Augmented Generation) framework, the chatbot can often rely on three knowledge bases to generate more relevant and actionable information for decision-makers, accelerating adoption.
Currently, Ved is running in BETA across various environments, and we’re excited to present several use cases that demonstrate its potential to drive faster adoption and usage.
Enhancing SRE Efficiency with Ved
Challenges SREs Face:
Site Reliability Engineers (SREs) invest significant time and effort into crafting complex queries to monitor systems, diagnose issues, and maintain system reliability. This process demands deep technical expertise and proficiency with specific query languages and dashboard tools. During critical production issues, managing multiple tools, dashboards, and queries while collaborating across teams becomes even more challenging for the SREs, especially under time constraints and CXO-level escalations.
Ved’s Solution: SREs can now draft the query in natural language and have Ved compile the necessary information from different systems and demonstrate the information in the form of a chat session with even dynamic dashboards. This reduces cognitive load and allows SREs to focus on higher-value activities.
Use Cases for SREs:
- Real-time Incident Analysis: “Show me the error rates for the payment gateway over the last two days.”
- Root Cause Analysis: “What caused the failure rates in payments and was there a correlation to a spike in CPU usage in our production servers last night?”
Impact on Business Operations
Need for Speed in Business Operations
Challenges in Business Operations: Business Operations teams often struggle with delays in accessing critical operational data. They rely heavily on batch reports generated by Technology teams, which are typically based on data extracted from database tables. This process is not only time-consuming but also results in outdated information by the time it reaches decision-makers. Additionally, the lack of real-time visibility into transactional data, such as success rates, failed transactions, and partner performance, hinders the ability to respond swiftly to emerging issues. These delays can lead to missed opportunities, slower response times to customer needs, and inefficiencies in managing partner relationships.
Ved’s Solution: With Ved, business teams can directly query real-time data, including logs and observability metrics, providing them with up-to-the-minute insights into operational metrics, partner performance, and system health. This immediate access empowers teams to make faster, more informed decisions, improving overall business agility.
Enhanced Real-Time Insights: Unlike traditional BI reports, Ved provides a real-time view into the entire data landscape, including success, failed, and broken transactions, offering richer business insights and enabling teams to monitor and optimize partner performance in real time.
Use Cases for Business Operations:
- Operational Efficiency: “What was the average TAT for Instant Payment transactions in the past two days?”
- Customer Experience Monitoring: “How many technical declines occurred in the last month?”
Sometimes, insights from an observability platform can provide significant value to a business function, even if we don’t initially realize it. Take, for example, a relationship or account manager and even field sales people and business correspondents. They often collaborate with customers or partners to ensure a superior experience and success. To do this effectively, they need real-time information on how the services provided to the customer are performing so they can address any issues proactively. Here are some examples of how they can interact with Ved to have these insights.
Use Cases for Relationship Managers:
- Client Transaction Analysis: “Show me the top merchants by Instant Payment transaction volume for this month.”
- Issue Resolution: “Were there any failed transactions for the IndigoArt merchant on June 15, 2024?”
- High-Value Transaction Monitoring: “Show me all the Instant Payment transactions that failed with an amount of more than Rs. 20,000 this year.”
Advanced Capabilities with Ved
Ved’s natural language processing enables users to ask complex, context-rich questions that were previously difficult to formulate without deep technical expertise:
- For SREs: “Which components had the highest alerts in the past three months?”
- For Business Operations: “What is the share of Instant Payment transaction amounts in business vs non-business hours yesterday?”
- For Relationship Managers: “Compare the error rate of Instant Payment transactions between the first and last week of the past month.”
Conclusion
This blog offers a preview of Ved’s capabilities, showing how it will assist Site Reliability Engineers (SREs) and Business Operations teams in powerful new ways. However, when it comes to monitoring critical production systems, precision is non-negotiable. Inaccurate alerts and actions can lead to significant disruptions, especially during late-night incidents, where a simple “model error” isn’t an acceptable excuse.
While we are excited about AI’s potential to revolutionize monitoring, we recognize that its initial impact will likely involve workflows where human expertise remains essential, with the nuanced understanding that SREs and business operations bring—the “tribal knowledge” of systems — will still be vital.
As we continue to advance Ved, our focus is on ensuring it enhances human capabilities and bridges the gap between observability and intelligent recommendations – but this journey requires careful development to maintain the trust and reliability that production environments demand. We look forward to sharing more on how Ved evolves and the practical applications it will enable in the future.
