High impact, insight driven survey experiences using Amazon Bedrock: Service Monitor’s adaptive survey intelligence system.
Overview
Service Monitor, a UK market research agency offering data insights to the hospitality, leisure, and retail sectors, required a partner to build a Generative AI powered review collection product (activeinsight.ai). Lambert Labs designed and delivered a survey intelligence system using Amazon Bedrock. The system dynamically tailors survey questions, analyzes free-text responses in real time, and surfaces actionable insights that are difficult to capture through static surveys. Built on Amazon ECS, the solution provides high availability, fault tolerance, and scalable operation of generative AI workloads.
Lambert Labs were highly professional throughout,
and took a very modern approach to cloud engineering
Jack Buesnel, Managing Director, Service Monitor
Opportunity / Customer Challenge
Service Monitor identified the need to provide timely, high-confidence insights to clients operating in fast-moving and customer-facing environments across hospitality, leisure and retail. Static consumer review surveys with fixed question sets could not adapt to individual experiences in real time, as follow-up questions had to be anticipated in advance rather than tailored to guest responses.
An AI powered system could edit the survey questions in real-time, thus eliminating redundant questions by understanding what has already been discussed. If a guest mentions the food was excellent in free-text, the system skips the food rating question. This makes surveys feel more tailored to an individual’s experience, and reduces respondent fatigue – driving higher completion rates.
An intelligent system can identify priority issues automatically – such as safety or hygiene – at an early stage, which will drive action from managers or stakeholders. There is also the opportunity to make use of historical feedback and location-specific context to tailor survey questions, and improve the quality of insights delivered to clients.
Solution
Lambert Labs engineered an adaptive survey intelligence system powered by Anthropic Claude and Amazon Nova Pro via Bedrock, which enables surveys to dynamically respond to customer input in real time. Survey questions are adjusted or added based on respondent sentiment, topic coverage and confidence. This reveals deeper and more relevant feedback without increasing the survey length, and avoids repeating similar questions.
Free text questions are analyzed on submission, and are used to surface actionable insights and improvement points to relevant stakeholders. The solution supports simultaneous model use and selection through the Amazon Bedrock Marketplace. Model invocation logs in Bedrock are used to perform data-driven evaluation of prompts and models, supporting ongoing and selection of the most effective LLMs for survey intelligence.
The platform is deployed using containerized workloads on ECS using AWS Fargate, with multiple Availability Zones to support scalability and resilience. Amazon CloudFront and AWS WAF provide availability and edge security, while private networking, monitoring and backup ensure enterprise grade security and reliability. Amazon CloudWatch Logs and Dashboards provide visibility into workload health alongside platform usage. Infrastructure is managed using Hashicorp Terraform, with automated CI/CD pipelines built on GitHub Actions to enable rapid iteration, testing, and deployment across development, staging, and production environments
Lambert Labs were highly professional throughout the project and took a very modern approach to cloud engineering. They have a deep knowledge of AWS and how to build solutions that are efficient, secure, and easy to manage. Communication was clear and well documented, and they adapted smoothly as requirements evolved. Overall, the experience was very positive, and I would be happy to work with them again. (Jack Buesnel, Managing Director, Service Monitor)
Outcome
Service Monitor can now deploy intelligent, contextually-aware surveys at scale – each one adapts in real-time to what customers say, asks relevant follow-ups, skips redundant questions, and customizes itself to each business without manual configuration. The result: higher completion rates, deeper insights, and operational scalability. Learn more about the product here.
Our goal was to build a production-ready, Generative AI powered, consumer review collection platform for Service Monitor on AWS. The team have delivered! (George Lambert, Founder & CEO, Lambert Labs)
About Service Monitor
Service Monitor is a UK based market research agency that specializes in customer experience and feedback for the hospitality, leisure and retail industries.
The company has over two decades of experience providing data-driven insights through a blend of methodologies including Mystery Shopper programs, online surveys and analytics.
Service Monitor assists clients to drive customer loyalty, ensure consistent brand compliance, and encourage positive change by uncovering real stories and recommendations.