Source Global Research: Accelerating Research Intelligence with AI-Powered Insights

The snapshot

About Source Global Research

Source Global Research is a London-based research and strategy firm specialising in the global professional services sector. The company provides data-driven insights to help consulting and professional services firms refine strategies, understand market trends, and enhance positioning through market analysis, client and brand insights and thought leadership evaluation.

Operating in a highly competitive market where speed and accuracy directly impact client value, Source maintains cloud-based resources hosting applications, datastores, and analytics tools integral to their business operations.

The Challenge

Source recognised an opportunity to enhance their data analysis capabilities. The firm’s success depended on rapidly transforming raw data into actionable insights, and with AI technologies maturing, they saw potential to increase efficiency and scale without adding headcount.

Three areas stood out as candidates for AI enhancement:

1. Qualitative research analysis – for surveys requiring qualitative analysis, continuing advances in AI tools and capabilities offer new opportunities to maximise efficiency and depth in the extraction and identification of salient information.
2. Entity recognition – AI could make it easier to identify company names and key terms to build searchable databases.
3. Thought leadership evaluation – Source’s well-established methodology for assessing content quality was detailed and rigorous. By integrating AI, they aimed to enhance the speed and scalability of this high-value process.

Source saw the potential of automation and AI to increase both efficiency and effectiveness, enabling the team to focus more time on high-value analytical work.

The Solution

Devoteam partnered with Source to deliver AI transformation through AWS-funded programmes, combining technical implementation with strategic production planning through a carefully structured approach designed to prove value before full-scale deployment.

Phase 1: Proof of Concept – Building AI Capabilities (April-May 2025)

The transformation began with a focused six-week engagement developing and validating three core AI tools, each addressing specific areas for efficiency gains.

Qualitative Summarisation Engine
Leveraging AWS Bedrock’s large language models, Devoteam built a customisable summarisation tool analysing both individual survey responses and aggregated answer sets. The solution incorporated metadata capture alongside text summaries, enabling intelligent indexing and search capabilities, with metadata tag capture achieving 70%+ accuracy.

Named Entity Recognition Pipeline
The team developed a hybrid approach combining AWS Bedrock’s generative AI capabilities with serverless pre- and post-processing services. This dual-model strategy maximised extraction accuracy whilst providing flexibility to validate entities against whitelists and make amendments as needed. The solution significantly exceeded expectations, achieving above 90% extraction accuracy whilst maintaining the ability to persist validated results in Source’s databases.

Thought Leadership Evaluation Tool
This solution used Amazon Textract for document processing and AWS Bedrock to assess publications against Source’s established evaluation methodology. Four specialised AI agents scored content against Source’s proprietary analytical framework, replicating the expert judgement previously performed manually by the analytics team.

Throughout the proof of concept phase, all solutions were developed in Python notebooks, providing both versatility and a clear path to production. Training and accuracy datasets were carefully curated for each use case, with performance metrics captured at every stage.

Phase 2: Production Readiness Assessment (August-September 2025)

Following successful proof of concept validation, Source engaged Devoteam for a comprehensive production readiness assessment under the AWS RAPID programme – a structured framework for accelerating GenAI deployment. This three-week engagement evaluated the AI tools across critical operational dimensions:

• Usage Analysis and Model Assessment examined usage patterns, request frequencies, and current LLM provider costs to identify optimisation opportunities before production deployment.
• Technical Implementation Review assessed prompt engineering practices, agent frameworks, and function implementations.
• Operational Requirements Analysis evaluated governance structures, data storage and performance benchmarking against SLAs including latency, uptime, and output quality measures.
• Production Planning delivered a comprehensive implementation roadmap addressing identified gaps, with specific recommendations for successful production deployment at scale.

The Value Delivered

The engagement delivered measurable impact whilst establishing foundations for Source’s ongoing AI journey.

Proven Accuracy & Efficiency Gains
The proof of concept demonstrated that AWS AI services could meet Source’s stringent requirements. The Named Entity Recognition pipeline significantly exceeded expectations at above 90% accuracy. The Thought Leadership Evaluation Tool enhances the existing process through baseline scoring across thousands of documents, whilst still allowing time, care, and effort for manually reviewing featured Thought Leadership, significantly increasing document processing capacity without proportional increases in analyst hours.

Reduced Risk Through Phased Approach
Rather than committing to full-scale production immediately, Source validated technical feasibility and business value through focused experimentation. The subsequent RAPID assessment provided clear guidance on governance, performance, and operational requirements – building stakeholder confidence and positioning Source for successful deployment at scale.

Accessible Transformation
Strategic use of AWS funding programmes enabled Source to pursue ambitious AI transformation that might otherwise have required prohibitive capital investment.

Why Devoteam?