Automating Cinode's CV Matching with an AI-Powered E-mail Solution

In Brief

• Cinode sought to explore further how they could automate and streamline their consultant matching process using AI and Machine Learning while maintaining data control through a proof-of-concept project. 
• Our team developed an AI-powered e-mail processing system in the AWS platform to automatically match broker requests with consultants, proving to reduce manual workload and ensure full data control.

Their Challenge

Cinode had already explored AI by launching a solution to translate consultant profiles. Now, they sought to explore further how they could automate and streamline their consultant matching process using AI and Machine Learning – without compromising data security.  

"Cinode is currently in a phase, like many others, of discovering how AI can help us stand out from competitors and stay ahead. After experimenting with OpenAI for a while, we wanted to explore and compare it with Amazon's suite of solutions. Therefore, we needed a partner who could provide us with AWS expertise, a pragmatic approach, and ease of collaboration. Naturally, our choice fell on Redeploy."

– Mark Kantor, Chief Product Officer at Cinode

Our Solution

Our team began by indexing all consultant profiles and developing two parallel tracks: a chatbot and an e-mail solution. The chatbot enabled customers to inquire about suitable consultants for specific assignments. The e-mail solution automatically and semantically processed requests from consulting brokers, matching them with consultant CVs from Cinode's database. The e-mail solution proved more functional than the chatbot, thanks to its advanced semantic capabilities.

In the e-mail solution we implemented several key components, leveraging AWS services, to maintain data integrity and security within Cinode's environment, including:

• Amazon Bedrock: Providing AI and ML capabilities in a controlled environment, Bedrock ensures secure and compliant AI processes aligned with Cinode’s data policies. It prioritizes privacy by avoiding the use of customer data for training and securely housing all data within AWS infrastructure.
• OpenSearch: Used as a vector database, OpenSearch allows Cinode to securely store, index, and match consultant profiles and broker requests, enabling efficient and secure data retrieval.
• Claude (Cohere model within AWS): For semantic email parsing and creation of vector embeddings.
• AWS Lambda: Automating the email matching process within a serverless environment ensures that data processing is securely handled within Cinode's infrastructure, reducing exposure to external systems. This approach also offers significant cost savings, particularly during periods of low traffic.
• SES (Simple Email Service): This service handles the reception and formatting of emails without needing a third-party email server, ensuring that all email data remains within AWS’s secure environment.
• Terraform: Used for infrastructure management, ensuring that all resources are provisioned and maintained consistently and controlled.

Using these AWS services ensured that all data handling, from e-mail reception to AI processing and data storage, occurred within a secure, compliant, and controlled environment, giving Cinode complete control over their data.

The Results

Our AI-powered email solution successfully automated the reading and understanding of broker requests, streamlining the matching process and reducing manual workload while giving Cinode complete control over their data.

Furthermore, the project facilitated Cinode in effectively testing AWS AI technology within their operational domain and has paved the way for future initiatives involving Bedrock and the development of data-secure AI—a collaboration we continue to support.