• Tibco’s ModelOps will help businesses ensure that the value of ‘the AI Model’ can be retained and improved throughout the model’s lifecycle.
• The complexity of AI can slow the deployment of models. Tibco’s ModelOps allows even non-programmers to move a model from development to deployment.
Artificial intelligence (AI) has quickly become a must-have tool for businesses. While AI has enabled businesses to gain new insights into their performance, its management and operational aspects have lagged behind, creating bottlenecks in deployment. In addition, companies are carrying a large amount of under-deployed and unrefreshed models that drain budgets and add to what is often referred to as ‘model debt.’ Some of these AI models are not utilizing the most recent data or have drifted away from their original business purpose. AI models and data sets continue to grow in size and complexity, requiring larger teams of software engineers to transform a model into an application.
Unfortunately, developer teams’ abilities to work on large data sets can be constrained by platforms, and developers located in different locations can make it difficult to coordinate workflows. Creating an AI model is a bespoke process adding to the development time as key functions such as pipelines are handcrafted or scripted.
Due to their specific purpose, AI models require extensive management tools. To help businesses manage, scale, and operationalize their AI assets, vendors are offering ModelOps. ModelOps focuses on getting AI models through validation, testing, and deployment phases quickly while maintaining model quality. It also focuses on ongoing monitoring, retraining, and governance to ensure peak performance and transparent decision-making. In helping businesses take advantage of the economic benefits, ModelOps also help ensure that all models running in production are aligned with stated technical and business KPIs.
Tibco is the latest vendor to offer a ModelOps solution for businesses. Developed in 2021 and released for general availability in July 2022, Tibco’s ModelOps solution offers a variety of management capabilities including lifecycle management, machine learning decision models, continuous integration/delivery, and champion-challenger/testing. In addition, Tibco’s ModelOps solution can integrate components that have been used in prior models, helping to reduce time spent on creating a bespoke solution.
The Tibco solution is format-agnostic and supports API-based models in the cloud or on-premises. Using its set of low-code tools, Tibco ModelOps enables any authorized business user, data analyst, or IT user to manage and deploy models into production. Users can also monitor model performance by utilizing customizable dashboards. Besides Tibco’s Spotfire application, ModelOps is compatible with non-Tibco platforms such as Tableau and Microsoft’s Power BI.
Tibco’s ModelOps solution will help customers reduce deployment and development time and better manage their AI models. However, businesses will also need to adapt in order to take advantage of the capabilities of their ModelOps platform. Accountability will be essential. One goal of ModelOps is to streamline the model production; a committee will need to be established to review rules implemented for new applications. Given the costs of building out new models, the ability to calculate costs and ROI will be crucial in order to ensure the value and costs are in line with budgeted resources.