Frequently Asked Questions
3.1. What is the recommended deployment strategy for large-scale enterprises with multiple continents, 100’s projects, and teams?
We recommend deploying Edmund AI for every continent where operations are active. This strategy helps to reduce installation costs, minimize latency, and facilitate the sharing of knowledge and resources among team members. Grouping Edmund instances will save deployment cost and achieve better hardware utilization.
3.2. Which is recommended for on-demand installation: a private cloud or virtual machines (VMs)?
We recommend using a private cloud over standalone virtual machines for on-demand installations. As we continuously develop new features, system requirements might change. A private cloud offers the flexibility to dynamically increase resources in a short time, ensuring that Edmund AI can adapt to evolving demands and maintain optimal performance.
3.3. Do hardware requirements for Edmund AI depend on the scale of data and organizational structure?
Yes, hardware requirements for Edmund AI are directly influenced by the volume of document data, the number of organizations, teams, and projects, and the overall number of uploaded documents. Larger datasets and more complex organizational structures necessitate greater hardware resources to ensure efficient processing and performance.
3.4. Are there limitations on the number of concurrent chat sessions Edmund AI can support?
Yes, Edmund AI is resource-intensive. If sufficient resources are not available to handle new requests, incoming conversations will be queued and will wait until existing conversations are completed and resources become available. This might occur during peak hours.
If it happens too often, your organization will be asked to increase its resources in order to ensure that the users have the best experience.
3.5. Can an organization manage all documents within a single project?
Technically, it is possible for an organization to store all documents within a single project. However, a more granular approach is highly recommended. Keeping projects granular helps achieve optimal results. While Edmund AI is effective at filtering most non-relevant data, processing a very large, undifferentiated dataset can increase processing time and potentially decrease the quality of the results.
3.6. How are data migrations handled between cloud/public deployments and on-premise/VM installations?
Currently, data migrations between shared cloud/public deployments and on-premise/VM installations are not fully automated. All documents and maintenance journal entries must be migrated in a semi-automated manner, which requires data re-indexing. Please note that chat history will be lost, and all users will need to recreate their accounts. Future enterprise support for automated migration might be added if there is sufficient demand.
3.7. What is the release frequency and standard release process for Edmund AI?
We are actively developing new features and rolling out updates on a weekly schedule. We aim for zero-downtime updates, but certain functionalities like document indexing or Edmund chat might experience temporary slowdowns during these periods. In some cases, zero-downtime updates may not be possible, and we allocate specific maintenance windows for such releases to minimize disruption.