Case Study: Utilising Language Recognition, Speed, and Enhanced Security to Make Social Media a Force for Good

Here, we take a detailed look at how the Cloud Combinator team collaborated with another cutting-edge AI service provider that provides intelligent systems to “make social media more social” for brands and users alike.

About Arwen AI

Arwen AI is a UK-based startup specialising in AI solutions to manage and enhance brands’ social media interactions. Founded in 2020 by Matt McGrory, Dr. David Cole, and Joel Bailey, Arwen.AI focuses on using AI to automatically detect and remove spam, toxic comments, and other unwanted content from social media platforms.

The team at Arwen have three core products. ‘Moderate’ is focused on identifying and removing toxic content from social media channels. ‘Engage’ helps brands identify and engage with meaningful conversations on social media, and ‘Customize’ allows brands to apply bespoke algorithms to their channels - creating an even more effective moderation and engagement. 

Arwen is trusted by some of the world’s most recognisable and innovative companies - including the AMG Petronas Formula One Team, Oracle Red Bull Racing, Oanda, ITV, Sports Illustrated, Fulham FC, and many more. In total, their clients have over 430 million followers - and Arwen’s platform ensures these followers are not exposed to racism, sexism, homophobia, bullying, abuse or severe profanity - nor are they spammed or click-baited through any of their partner brands’ social media accounts.

Customer Challenge

Having recently secured the investment needed to push their platform to the next level, the Arwen team identified a series of challenges they felt Cloud Combinator could help with. 

Central to their platform is the need to accurately identify and classify diverse content types - ranging from toxic comments to specific complaints. Accuracy in this area was crucial - not only for improving user experience but also for maintaining the integrity and reliability of the platform. However, the existing systems occasionally struggled with higher rates of false positives and negatives - issues that could potentially lead to user dissatisfaction.

Alongside accuracy, Arwen recognised that their users required a service capable of handling their social media needs with as little lag as possible - especially during peak engagement periods. Behind the scenes, this meant ensuring infrastructure was capable of handling the escalating data volumes that came hand-in-hand with an ever-growing user base and and the increased possibility of peak times when multiple customers would experience simultaneous spikes in comments - for example, when an F1 Grand Prix coincided with a Premier League football match and a major TV programme being broadcast.

Of course, none of these service enhancements could come at the expense of security. With Arwen’s extensive list of client companies extending around the world, every element of every solution had to comply with stringent global and regional data protection regulations. What’s more, each of these enhancements and the security requirements that went with them had to be integrated seamlessly into the company’s existing infrastructure without any risk of disrupting service continuity.

Partner Solution

To address Arwen’s challenges, the team at Cloud Combinator used our AI and ML Project Accelerator process to create a solution that would enhance text classification accuracy, ensure scalability, and comply with stringent security standards. 

Key Components of the Solution:

• AWS Bedrock Integration: This essential component allowed for the deployment and management of Large Language Models (LLMs) specifically configured for Arwen AI’s text classification needs. AWS Bedrock provided the flexibility to customise and optimise models to accurately detect nuanced social media content, including toxic comments and specific complaints.
• Custom Model Development: Tailored machine learning models were developed to specifically address the unique challenges of Arwen AI’s content classification. These models were fine-tuned to improve precision in identifying various types of content, thereby reducing false positives and enhancing user trust.
• AWS Lambda for Model Invocation: Utilising AWS Lambda enabled serverless computing that handled high volumes of data requests dynamically. This architecture was crucial for maintaining system responsiveness and efficiency, particularly during periods of high data throughput.
• API Gateway Configuration: The implementation of an API Gateway streamlined the flow of data between Arwen AI’s application layer and the AWS-hosted backend. This setup not only improved data transfer reliability but also enhanced the overall speed and efficiency of the system.
• Security and Compliance Measures: Arwen's security was enhanced by implementing Virtual Private Cloud (VPC) to VPC connectivity, ensuring all operations occurred within a protected environment. This setup, including data encryption and secure API endpoints, minimised vulnerabilities, maintaining data integrity and compliance with international regulations - while safeguarding user data confidentiality.

With service disruption simply not an option, the integration of AWS Bedrock into Arwen AI’s existing technology stack was meticulously planned to ensure no downtime. This involved extensive architectural redesigns and the development of new interfaces to accommodate the advanced LLM capabilities. To guarantee seamless service continuity during the upgrade, deployments were carefully staged, and, in some cases, parallel systems were maintained temporarily.

Results and Benefits

The implementation of AWS Bedrock and advanced machine learning models by the Cloud Combinator team significantly enhanced Arwen AI’s social media account monitoring capabilities, producing measurable improvements in both accuracy and operational efficiency.

Enhanced Classification Accuracy: The introduction of a bespoke classifier for detecting complaints led to a significant reduction in false positives and negatives, sharpening the precision of content categorisation. This was a significant achievement, as Arwen Co-founder and CTO David Cole puts explains:

Any single percentage improvement above our existing 96% is really difficult. These are the “hard yards” of our work, where we’re tackling very sophisticated bots who are themselves using AI.

Alongside this, enhancements in filtering out toxic content bolstered the safety and quality of online interactions. These refinements have allowed Arwen AI to adapt dynamically to the shifting language and nuances of social media, resulting in this significant leap in precision compared to previous systems.

Increased Operational Efficiency: AWS Lambda's serverless architecture played an important role in accelerating the text classification process, essential for real-time monitoring. This element of the project has ensured that Arwen AI can handle larger data volumes swiftly and efficiently, significantly reducing response times - even during peak data influx periods. This capability is crucial for maintaining high service levels without compromising on performance - therefore enhancing overall user satisfaction and ongoing trust in the platform.

Why Cloud Combinator?

Driven by our unique AI and ML Project Accelerator process, the Cloud Combinator team’s track record around optimising AI and ML workflows is second-to-none. As such, we were a perfect fit for a fellow AI company looking to enhance the systems that power their cutting-edge offerings - especially given that our Advanced AWS Partner status means can leverage funding through Amazon to keep projects cost-neutral. 

With experience of successfully delivering projects in fields as diverse as global recruitment and round-the-clock manufacture, the team at Arwen was also absolutely confident that avoiding service disruption is a non-negotiable for Cloud Combinator - with our first-hand application of tried-and-tested strategies that ensure zero downtime for mission-critical systems. This meant Cloud Combinator’s solution wasn’t just effective; it could be rolled out seamlessly - while always working within all global data security regulations.

Client Satisfaction

“The applications of this work will have a wide-ranging impact, not just around classification but also in the customisation we can offer our customers. Our partnership with Cloud Combinator has been pivotal in this. Their AI knowledge and expertise was invaluable for model selection and prompt creation to accelerate our LLM journey. Also the whole process was smoothly managed with excellent communication throughout.” 

David Cole, Co-founder and CTO, Arwen AI