The Impact
~45-60 min → ~10-15 min (Est.)
20-30 min → <2 min (Est.)
15-20 hrs/week (Est.)
Up to 80% (Est.)
About the Customer
REGION
INDUSTRY
JUMP TO SECTION
Share this Article
Executive Summary
A UK-based field services organization engaged Matellio to automate two manual, time-intensive workflows on its existing operations platform: field survey data capture and contract change analysis. Surveyors previously spent significant time completing lengthy digital forms on-site, often in poor-connectivity conditions that forced paper-based workarounds and double data entry, while office teams manually compared original contracts against updated survey information to identify changes, calculate financial impact, and notify stakeholders. Matellio built and deployed two production generative AI capabilities on Amazon Bedrock — a Voice-to-Form AI Survey Assistant and an AI Contract Change Analyzer — integrated directly into the customer’s existing platform.
Customer Challenge
The customer’s field survey and contract change processes relied heavily on manual work. Surveyors spent significant time completing lengthy digital forms while working on-site, and poor connectivity in the field created paper-based workarounds, double data entry, and delays. Once a survey was submitted, teams also had to manually compare the original contract against updated survey information line by line to identify changes, understand the reasons behind them, calculate the financial impact, and notify relevant stakeholders.
Left unaddressed, these manual processes would come under increasing pressure as survey volumes grew, limiting the customer’s ability to improve the efficiency of its field-to-office workflow without continuing to add manual effort.
Goals and Objectives
Why AWS
The customer chose AWS because the proposed generative AI solution could be built entirely on AWS-native AI and cloud services, including Amazon Bedrock, Amazon Transcribe, Amazon S3, AWS AgentCore Runtime, AWS Lambda, Amazon API Gateway, and Amazon CloudWatch. Amazon Bedrock provides the Claude generative AI capability required for natural-language understanding and contract analysis, while Amazon Transcribe supports the voice-to-text workflow for field surveys. The solution was also designed to operate in the UK AWS region, supporting the customer’s UK GDPR requirements and keeping the solution within its existing cloud environment.
The customer chose Matellio because of the existing relationship and Matellio’s direct knowledge of its operations platform. Matellio had previously delivered a CRM transformation project for the customer and had built the platform that these new AI capabilities would integrate with. This existing understanding of the customer’s workflows and technology provided a natural foundation for extending the platform with generative AI capabilities, rather than introducing a separate platform or requiring the customer to start with a new technology partner.
Partner Solution
Matellio built and deployed two production generative AI capabilities integrated with the customer’s existing operations platform, automating two previously manual operational workflows: field survey data capture and contract change analysis.
A mobile-based workflow that enables surveyors to record observations by voice while working on-site, including in low- or no-connectivity environments. Amazon Transcribe converts the recordings to text, while Amazon Bedrock with Claude interprets the transcript and maps observations to the appropriate survey fields. Surveyors review and correct the AI-generated information before submission. The application securely stores inputs locally when offline and automatically synchronizes them with the platform once connectivity is restored.
A production AI workflow that automatically compares the original contract with updated survey information. It identifies additions, removals, substitutions, and other line-item changes, incorporates relevant context from survey notes, calculates the financial impact, and generates a structured summary. Results are surfaced within the platform and automatically emailed to relevant stakeholders.
Solution Architecture
The solution is implemented using Amazon Bedrock (Claude), Amazon Transcribe, Amazon S3, AWS AgentCore Runtime, AWS Lambda, Amazon API Gateway, and Amazon CloudWatch, providing the AI reasoning, speech-to-text, storage, agent runtime, serverless processing, API integration, and monitoring capabilities the solution requires.
Amazon Bedrock was used to build and manage the generative AI models, avoiding the need to provision and manage GPU-backed model-serving infrastructure directly. Amazon Transcribe was used for speech-to-text rather than a custom or open-source speech recognition pipeline, given its native integration with other AWS services and its ability to handle field-recorded audio without additional infrastructure to build or maintain.
The solution’s use of managed AWS services – Amazon Bedrock, Amazon Transcribe, AWS Lambda, and Amazon API Gateway – allowed both new capabilities to be delivered as additions to the customer’s existing operations platform rather than as a separate system, minimizing operational overhead and keeping the solution within the customer’s existing AWS footprint and UK data residency requirements.
Results and Benefits
Post-launch, Matellio applies its standard continuous improvement practice to this solution: feedback on survey-mapping accuracy and contract-change output is captured, evaluated, and used to refine prompts and field mappings on an ongoing basis, consistent with the feedback-and-improvement loop described in Matellio’s generative AI methodology and production support practice.
Let’s turn your next big idea into a scalable product, engineered for speed, reliability, and real-world impact.
