Case Studies | SaaS B2B

Sharper detection of toxic content

Content Analysis using LLM and Comprehend

About

A SaaS B2B business working with Cloud Combinator on AWS. The client is anonymised at their request.

Challenge

The challenge had three focus areas.

Reducing reliance on third-party classifiers

The client depended on external classification services. It wanted its own model that it could control, tune and run cost-effectively, reducing that dependence without losing accuracy.

Detecting more, more accurately

The new classifier needed higher accuracy than the existing one, a new complaint category alongside toxicity, and the ability to infer patterns from raw data and draw conclusions from the analysis.

Proving the right AWS platform

Before committing, the client needed to validate whether Amazon Bedrock, and for narrow classification Amazon Comprehend, was the right foundation, with the option to fine-tune if base accuracy fell short.

Solution

We ran the engagement through our Cloud Accelerator, moving from understanding the client's product to a tested pipeline in their own AWS account.

1,000

Labelled records used to train the custom classifier

2

AI approaches built and evaluated, Bedrock and Comprehend

~$1,365

Projected monthly model run cost (projected)

By the numbers:

  • 1,000 - Labelled records used to train the custom classifier
  • 2 - AI approaches built and evaluated, Bedrock and Comprehend
  • ~$1,365 - Projected monthly model run cost (projected)
Changes

Both a LLM approach on Amazon Bedrock and a custom Amazon Comprehend classifier were built and tested against the client's labelled data, with results captured across toxicity and sentiment categories. The work gave the client a clear, evidence-based route to bringing classification in-house on AWS.

  • A path off third-party classifiers — The client gained a route to run classification in its own AWS account, reducing dependence on external providers.
  • New complaint detection — The work extended classification beyond toxicity to a new complaint category, widening what the client can flag.
  • Two engines, measured — Results were produced for both the Bedrock LLM and the custom Comprehend classifier, so the platform choice rested on evidence.
  • Fine-tuning option retained — The design allowed the model to be fine-tuned on the client's data where base accuracy fell short.
  • Handover — Knowledge shared through the accelerator immersion left the client's data science team able to keep developing the models.

With a tested, AWS-native classification pipeline and a clear read on Bedrock and Comprehend, the client is positioned to bring more of its content analysis in-house and extend it to new categories as its platform grows.

AWS Stack

Amazon Bedrock

For large language model classification of social media content.

Amazon Comprehend

For a custom-trained text classification model.

Amazon S3

For secure storage of labelled training and test data.

AWS Lambda

For serverless model invocation within the pipeline.

Amazon API Gateway

For accepting content from the client's application.

YOU MIGHT LIKE

Related success stories

View all case studies

Case Studies | Insights

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.
  • 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.
Read more
CONTACT US

Ready to turn AI into impact?

We'll help you spot the highest-value opportunities, reduce risk around your first AI initiative, and define a clear path to results from day one.

Why talk to us:

Outcome-driven recommendations

AWS-recognised delivery expertise

Risk-aware AI adoption

Clear next step, not a sales pitch

Start with a focused 20-minute conversation about your goals — no pressure, no commitment.

We use cookies

This website uses cookies to enhance user experience and to analyze performance and traffic on our website.

See our Privacy Policy for details.