Introduction
Managing cloud costs is a complex challenge for many organizations. Anodot, a leader in autonomous business monitoring, addresses this with its innovative OpenAI-based solution, CostGPT. Partnering with Automat-it and leveraging their POC recommendations, Anodot moved an agent to Amazon Bedrock from OpenAI and achieved a 30% classification performance improvement.
Conversational analytics: simplifying cloud cost management
CostGPT uses Natural Language Processing (NLP) to simplify cloud cost management through conversational analytics. Users intuitively interact with their cloud cost data, asking questions in natural language to retrieve detailed insights and take informed actions quickly.
Autonomous cloud cost Intelligence model
CostGPT’s core is Anodot’s autonomous cloud cost intelligence model, which combines advanced analytics, machine learning, and automation to provide comprehensive insights into cloud spending. Key features include:
- Dashboards and reports: tailored visualizations presenting cost data in an easily digestible format.
- Anomalies and forecasts: detection of cost anomalies and accurate forecasting to predict future spending trends.
Amazon Bedrock: a fully managed gateway for LLMs
Amazon Bedrock provides a suite of foundational models for machine learning (ML) and artificial intelligence (AI) applications. Designed to simplify the deployment and scaling of ML models, Bedrock offers pre-trained models that can be customized with minimal coding. Key benefits include reduced time to market for AI solutions, cost-effective scaling of ML workloads, and seamless integration with AWS services.
Technological innovations and solutions
CostGPT is built on a robust technological foundation designed to enhance performance and deliver accurate results. Key technological highlights include:
- Rich context with vectors: utilizing vectors like FAISS, SVM, and kNN for embeddings to provide rich contextual information.
- Efficient similarity search and clustering: implementing techniques like BM25 and TF-IDF for similarity search and clustering.
- Classification with confidence: using log probabilities to classify answers with confidence, ensuring users receive reliable information.
- Advanced caching mechanisms: implementing various caching strategies to improve response times and ensure relevant data retrieval.
Outcome
CostGPT 30% performance improvement in classification
The Amazon Bedrock-based classification agent, co-developed with Automat-it, addresses the complexity of cloud cost management by providing straightforward answers to complex queries, such as:
- Listing services launched within a specific timeframe.
- Calculating cost savings from optimization efforts across different regions.
- Determining the percentage coverage of various EBS storage types.
Identifying services most impacted by anomalies and accounts with significant cost increases.
Amazon Bedrock’s flexibility, scalability, and simplicity enabled Anodot to test alternatives to GPT-4. Benchmarking and evaluation led to the selection of Claude 2 for usage with one of the agents, which proved to be 30% faster while maintaining similar performance levels to GPT-4.
Why customers choose Amazon Bedrock
Amazon Bedrock offers several advantages over relying solely on GPT-4:
- Model diversity: access to a variety of foundational models optimized for different tasks.
- Customization: easier customization of pre-trained models for better performance and relevant outputs.
- Cost efficiency: flexible model selection aligning with budget and performance needs.
- Integration with AWS Services: seamless integration facilitating scalable AI development.
- Scalability and reliability: leveraging AWS’s robust infrastructure for scalable and reliable AI solutions.
- Security and compliance: secure environment with comprehensive compliance certifications.
- Innovation and updates: access to the latest advancements in AI and ML.
The future of CostGPT
Anodot continues to innovate and expand CostGPT’s capabilities, focusing on improved latency, accuracy, and cost efficiency. Future enhancements include supporting complex flows, trend prediction, right-sizing recommendations, and integrating incident management with cost and usage data.
“Our CostGPT solution is revolutionizing the way companies approach and control cloud cost management. The integration with Amazon Bedrock has not only enhanced the performance and scalability of our solution, but also allowed us to provide our users with faster insights. Our collaboration with Automat-it has been pivotal in delivering a robust tool that empowers businesses to manage their cloud expenses more efficiently,” said Alexander Shereshevsky, Machine Learning Architect at Anodot.
Conclusion
CostGPT by Anodot revolutionizes cloud cost management by making it more accessible and efficient through conversational analytics. Combining advanced technology with user-friendly interfaces, CostGPT empowers organizations to gain deeper insights into their cloud spending and proactively optimize costs.
As Anodot enhances this platform, the future looks promising for organizations navigating the complexities of cloud cost management. Automat-it will continue collaborating with Anodot to improve performance and reduce costs.
