## Deployment FAQs Deploy algoQA in your own cloud infrastructure within your firewall, use the cloud instance hosted by AlgoShack, or run it on-premises to match your infrastructure, security, and testing requirements. You can deploy in your own cloud infrastructure within your firewall, or you can use the cloud instance hosted by AlgoShack. The platform also supports on-premises deployment. On-premises deployment is supported for algoQA, but with specific considerations and limitations: 1. Deployment Models: - SaaS Deployment (Recommended): Hosted and fully managed on AlgoShack's secure AWS infrastructure with GenAI features working out-of-the-box. - On-Premises Deployment: Customers can host algoQA in their own infrastructure. GenAI support is limited unless specific network and model compatibility conditions are met. Default on-prem installs disable GenAI unless network access to the AWS Bedrock domain is allowed. 2. Options to have GenAI features for on-premises Deployment: - algoQA Bedrock Domain Whitelisting: Customer firewall or VPC must allow outbound HTTPS access to AWS Bedrock endpoints. - Provide Matching AWS Infra: A dedicated Bedrock-enabled AWS account/environment must be configured and accessible to algoQA. - Customer-Hosted Equivalent Models (Unsupported): While some customers may have identical models, AlgoShack does not support deploying or maintaining customer-hosted LLMs due to variability in model quality, runtime behavior, quantization, API compatibility, etc. 3. Key Considerations: - algoQA does not commit to deploying or managing LLMs on customer infrastructure. - If GenAI access is unavailable, relevant features degrade gracefully or remain disabled, without affecting core algoQA functionality. - Customer test data never leaves their infrastructure if using on-prem with Bedrock access. 4. On-Premises Deployment of algoScraper with GenAI Capabilities: - Deployment using AWS Bedrock: Requires whitelisting specific AWS Bedrock domains and ensuring HTTPS access on port 443. - Deployment without using AWS Bedrock: Requires availability of a 7B or above LLM variant and an API endpoint to access these models. Architecture: - LLM Provider: Amazon Bedrock - Model Usage: Uses foundation models via Amazon Bedrock APIs. - Data Handling: Stateless and transient prompts ensure zero data retention. - Frontend and API Gateway: The algoQA UI communicates with a secure backend, which mediates interactions with LLM services through authenticated Bedrock endpoints. Yes, scripts can be created centrally and executed in client environments. algoQA supports the creation of automation scripts in a centralized environment. These scripts can then be executed in various client environments, including on-premises and cloud deployments. This depends on the automation type and framework type you choose. Customers have the option to deploy algoQA within their own infrastructure and maintain full control over their environment. Please note: - GenAI capabilities are limited in on-premises environments unless specific network and model compatibility requirements are satisfied. - By default, GenAI features are disabled in on-prem installations. - To enable GenAI functionality, the deployment must allow secure network access to the AWS Bedrock domain and meet the required compatibility conditions. The scripts created using algoQA can be executed on the UAT environment without external dependencies, meaning no data needs to be transmitted to the environment for executing scripts. algoQA can be deployed on the on-prem UAT environment if scripts are to be generated and modified on the UAT environment. Otherwise, scripts can be transferred and executed. The size of the already generated scripts if only scripts are transferred. The size is dependent on the number of test cases, additional customization, etc. Scripts to be executed in the framework and language chosen. Prerequisites are dependent on the framework and language (chosen during configuration of the algoQA project) of the scripts. The detailed document is shared separately. No, Since the entire deployment resides within the client's environment and network, all data stays within the client's controlled environment. No data leaves the server once the on-premises deployment is in place. It can be deployed in-house within a firewall for privacy, with options for local or approved cloud LLMs (e.g., AWS foundational services), ensuring data privacy and organizational control. algoQA is a SaaS solution by default. For clients requiring on-premises deployment, the platform is deployed directly within the client network on one of the client's own servers, ensuring full control over the environment. All artifacts including prompts, screenshots, logs, and test artifacts are stored on the server hosting the algoQA solution within the client's on-premises environment. Nothing is stored externally. For on-premises deployments hosted on the client server, algoQA uses the client's approved LLM rather than any third-party model. The platform plugs into whichever LLM the client approves, ensuring compliance with internal AI policies. The algoQA platform and its browser extension can only be used with the client's public sites. Any tools must be approved by the client before they can be used. Whitelisting the IP address is necessary for security. It ensures that only authorized systems can install algoQA and its browser extension, helping to prevent unauthorized access and ensuring the tool is used in a secure environment. Yes. One of AlgoShack's radiology informatics customers uses Google Cloud to execute scripts from algoQA. For banks and medical-device organizations, the question isn't whether LLMs are useful but whether they can be used within security, privacy, and approval requirements. On-premises deployment keeps sensitive data and AI workloads within the organization's environment and firewall, reducing concerns about sending data to external services. As private AI becomes more accessible, on-premises deployment directly addresses the concerns slowing adoption in regulated enterprises.