AWS Only
Enhancing Iris’s Client Onboarding with LLM-Powered Service Recommendations
Iris partnered with Firemind to develop a Proof-of-Concept (POC) aimed at automating service recommendations for new clients using Large Language Models (LLMs). This solution streamlined the onboarding process by providing relevant recommendations based on historical data from similar clients.
AWS ONLY ·
Enhancing Iris’s Client Onboarding with LLM-Powered Service Recommendations
Iris partnered with Firemind to develop a Proof-of-Concept (POC) aimed at automating service recommendations for new clients using Large Language Models (LLMs). This solution streamlined the onboarding process by providing relevant recommendations based on historical data from similar clients.
At a glance
Iris provides essential software solutions for managing core business operations, with a focus on delivering accounting-related services and recommendations to support their clients' financial management needs.
Challenge
Iris needed to automate the generation of accurate service recommendations for new clients based on data from similar existing customers.
Solution
Firemind developed an LLM-powered system that extracts and analyses data from company reports, generating service recommendations using Amazon Web Services.
Services Used
Amazon Bedrock
AWS Step Functions
Amazon Athena
Amazon S3
Outcomes
Key information successfully extracted and structured from company reports.
LLMs delivered accurate service recommendations, aligned with Anazon Athena data.
Business challenges
Automating Accurate Service Recommendations for New Clients
Iris sought to enhance their client onboarding process by delivering more precise and relevant service recommendations to new clients. Their traditional approach involved a manual process that was time-intensive and did not fully leverage the extensive data available from existing clients. This presented an exciting opportunity for Iris to innovate and further streamline their operations.
Eager to improve efficiency and consistency, Iris aimed to automate the recommendation process. By doing so, they could ensure that new clients received tailored suggestions based on a thorough analysis of historical data, all while maintaining the high standards of service, this automation was seen as a key step towards boosting operational efficiency, speeding up the onboarding process, and enhancing client satisfaction.
"Firemind's PULSE tool was able to quickly and accurately summarise vast datasets of customer surveys, ensuring we could take action against any negative customer experience elements."
Dean Macfadyen, Data Platform Engineer — Generative AI Customer
Solution
Deploying LLMs to Streamline Service Recommendations
Firemind implemented a POC that utilised Large Language Models (LLMs) accessed through Amazon Bedrock. The solution involved creating a recommendation system that automatically generated service suggestions for new clients based on the services utilised by similar existing customers.
The system processed company reports in PDF format, extracting key data and converting it into structured CSV files, which were then uploaded into an Amazon Athena table. Using predefined SQL queries, relevant data was retrieved and analysed. The LLMs then processed this data to generate natural language responses tailored for Iris accountants, providing accurate and relevant service recommendations.
Firemind ensured that the LLMs stayed within the scope of the data provided, avoiding recommendations outside of the available information. This was achieved through careful prompt engineering and integration of various AWS services, resulting in a system that effectively supported Iris’s client onboarding process.
Increased efficiency in client onboarding:
The implementation of the LLM-powered system streamlined Iris’s client onboarding process, significantly reducing the time required to generate service recommendations. By automating data extraction and analysis, the system allowed accountants to focus on higher-value tasks, speeding up onboarding and improving overall operational efficiency.
Accurate and data-driven recommendations:
By analysing client data and ensuring recommendations stayed within the scope of services offered, the LLM-powered system provided highly accurate and relevant service suggestions. This not only increased the accuracy of recommendations but also enhanced client satisfaction.
Enhanced upselling opportunities:
The project significantly improved upselling opportunities for Iris accountants by automating the analysis of new customer profiles. The system generated real-time service recommendations, allowing accountants to instantly suggest additional services during the onboarding process, boosting both efficiency and revenue growth.
Model Spotlight
Claude Sonnet 3
We chose Claude Sonnet 3 for its ability to handle the initial testing and processing requirements of the Proof of Concept (PoC). By leveraging test data, Claude Sonnet 3 was able to quickly analyse the information available from the Amazon Athena queries and generate structured, relevant recommendations for upselling services.
This model played a crucial role in speeding up the manual task of analysing existing client portfolios to determine what services were being offered. Claude Sonnet 3 recommended services based on a new customer’s profile, such as revenue, location, and operating areas, providing real-time suggestions within minutes – far faster than traditional manual research processes.
This allowed Iris accountants to focus on upselling opportunities immediately after a new customer was onboarded, enhancing their ability to offer relevant services and streamline business processes.
Claude Sonnet 3.5
We chose Claude Sonnet 3.5 for its enhanced capability to understand complex contexts and deliver precise, relevant summaries, making it the ideal model for the final phase of the PoC.
Claude Sonnet 3.5 was especially effective in providing recommendations that closely aligned with the new customer’s profile, including key insights such as revenue and geographical data, while avoiding unnecessary or irrelevant suggestions.
This model eliminated the need for accountants to manually investigate potential upsell opportunities by delivering actionable recommendations instantly after the onboarding process. By automating this task, Claude Sonnet 3.5 reduced the time spent from days to minutes, empowering accountants to capitalise on upsell opportunities efficiently and consistently.
Why Firemind
Firemind’s track record of delivering tailored, innovative solutions that address complex business challenges made them the ideal partner for this project. Iris was particularly impressed by Firemind’s ability to seamlessly integrate generative AI technologies into existing workflows, ensuring a smooth transition from manual processes to an automated, data-driven approach.
1920%
Workflow speed increase
1,000 documents in 12.5 minutes compared to 4 hours by human operator
88%
Keyword accuracy
for leading categories within data modelling & training
Added value
We were also able to take all stakeholders on a journey concerning how Cloud Adoption and Machine Learning would benefit their customers and their business continuity, all in a 12 week timeline from start to finish.
Get in touch
Want to learn more?
Seen a specific case study or insight and want to learn more? Or thinking about your next project? Drop us a message below!