Client Name: Chatwit

Chatwit partnered with DevOps TechLab to enhance conversational business analytics with an AI-powered agentic investigation solution running on AWS.

The Challenge

Chatwit needed a more efficient way for users to investigate business and sales performance without depending on technical teams to manually prepare and execute database queries.

Traditional data analysis required users to identify relevant information, prepare SQL queries, run multiple queries, and interpret the results before reaching a conclusion.

The objective was to provide a conversational experience where users could describe a business concern in natural language and allow an AI agent to autonomously investigate the underlying data.

Key challenges included:

• Manual data investigation and analysis
• Dependency on technical resources for SQL queries
• Multiple queries required to investigate complex business questions
• Difficulty connecting findings across different data points
• Limited contextual follow-up during analysis
• Time required to move from a business question to actionable insight

Discovery

DevOps TechLab worked with Chatwit to understand its conversational analytics requirements and identify scenarios where an AI agent could perform autonomous, multi-step investigations.

The discovery process focused on understanding how users interact with business data and how the investigation process could be automated while maintaining controlled access to production databases.

The discovery included:

• Understanding conversational analytics requirements
• Identifying relevant business data and database structures
• Understanding relationships between available data
• Identifying complex, multi-step investigation scenarios
• Defining secure, read-only production database access
• Designing the agent’s planning, execution, reasoning, and recovery workflow
• Identifying opportunities to reduce manual data-analysis effort

Based on these requirements, DevOps TechLab designed the QueryDash Agentic Investigation Platform using Amazon Bedrock and Amazon Nova Pro.

Onboarding

DevOps TechLab onboarded Chatwit onto the QueryDash Agentic Investigation Platform, enabling users to investigate business performance through natural-language conversations.

A typical investigation can begin with a goal such as:

“Investigate why sales performance changed and guide me through the findings.”

The agent interprets the objective, determines what information is required, discovers relevant database structures, generates and executes validated SQL queries, analyzes the results, and determines the next investigation step.

Users can continue asking follow-up questions while the investigation context is maintained.

The onboarding included:

• Database connectivity configuration
• Amazon Bedrock and Amazon Nova Pro configuration
• Secure database credential management
• Read-only SQL validation
• Conversational investigation context
• Multi-step agentic investigation workflow
• Real-time response streaming
• AWS application deployment
• Validation of representative conversational investigation scenarios

Operations & Support

DevOps TechLab provides ongoing operational support for the Chatwit solution and its AWS environment.

The support model focuses on maintaining application availability, secure database access, and reliable AI-agent execution.

Operations and support activities include:

• Application and AWS infrastructure monitoring
• AI agent execution monitoring
• Database connectivity support
• Query execution and validation support
• Investigation workflow troubleshooting
• Production issue resolution
• Application logging and monitoring
• Secure credential and access management
• AWS infrastructure maintenance

Production database access is controlled through application-level validation. SQL generated by the AI agent is validated before execution, with write and schema-changing operations rejected.

Optimisation & Advisory

DevOps TechLab continuously works with Chatwit to improve the effectiveness, reliability, and scalability of its conversational AI investigation capabilities.

Optimisation and advisory activities include:

• Optimising conversational investigation workflows
• Improving agent planning and multi-step reasoning
• Improving SQL generation and validation
• Enhancing error recovery and replanning
• Improving contextual follow-up interactions
• Reviewing AWS infrastructure performance and utilisation
• Monitoring application and AI-agent performance
• Reviewing security and access controls
• Advising on additional conversational analytics scenarios
• Identifying opportunities for further AI-driven automation

The reusable architecture enables Chatwit to expand into additional business investigation scenarios without creating a separate analytics workflow for each requirement.

Summary

DevOps TechLab helped Chatwit introduce an AI-powered Conversational Investigation Agent that enables users to investigate business and sales performance through natural-language conversations.

Instead of requiring users to know SQL or rely on predefined dashboards, the agent can autonomously investigate a business objective through multiple steps.

The agent discovers relevant data, generates and validates queries, analyzes results, evaluates the evidence, recovers from failed queries, and continues the investigation based on the findings.

The conversational experience also allows users to ask follow-up questions while maintaining investigation context.

The solution combines Amazon Bedrock, Amazon Nova Pro, AWS compute, managed databases, conversational memory, real-time streaming, and controlled read-only data access to deliver a production-ready agentic analytics capability.

Key Benefits

• Conversational business investigation
• Autonomous multi-step data analysis
• Natural-language interaction
• Dynamic data discovery
• Context-aware follow-up questions
• Evidence-based findings
• Automatic recovery and replanning
• Real-time investigation results
• Controlled read-only production access
• Reduced dependency on manual SQL analysis

Note: If you publish quantitative outcomes such as the previously discussed ~60 minutes to ~5 minutes reduction, use those figures only if they are supported by Chatwit’s actual measurements or documented customer evidence.

About DevOps TechLab

DevOps TechLab is a cloud and DevOps technology company focused on helping businesses design, deploy, secure, and operate modern applications on cloud platforms.

Our expertise spans cloud infrastructure, DevOps automation, application modernisation, cloud security, managed services, Generative AI, and Agentic AI solutions.

By combining cloud engineering with AI capabilities, DevOps TechLab helps organisations automate manual processes, improve operational efficiency, and build production-ready intelligent applications on AWS.