AI “Council of Models” Improves Workflows and Outcomes
Selecting the best AI models for each step of a workflow while properly preparing structured and unstructured enterprise data enables a more effective systems engineering approach for biomanufacturing. The post AI “Council of Models” Improves Workflows and Outcomes appeared first on GEN - Genetic Engineering and Biotechnology News .
The AI industry has seen biopharmaceutical manufacturers adopt an end-to-end systems engineering approach to maximize the advantages of artificial intelligence (AI) in their workflows. Relying solely on one model often proves insufficient, especially as complexity escalates and therapeutics advance from pilot stages into production.
AI models differ in their capabilities, and even repeated runs of the same model can yield inconsistent outputs, a concern highlighted by Farshid Sabet, CEO and co-founder of Corvic AI. He warns that small inaccuracies can compound rapidly, leading to unreliable results in production environments. To tackle these challenges, Corvic AI recently evaluated leading frontier AI models using their workflow orchestration platform, focusing on their ability to transform piping and instrumentation diagrams (P&IDs) into XML files.
While frontier AI models excel at general text generation, engineering workflows introduce an entirely different level of complexity. Corvic AI addresses these issues by integrating and orchestrating multiple AI models, selecting the most suitable one for each workflow stage based on task, performance requirements, and cost. This approach, likened to a "council of models," amalgamates the strengths of various models to enhance accuracy, repeatability, and efficiency.
Corvic AI's platform emphasizes organizing enterprise knowledge through a semantic layer, allowing AI systems to grasp relationships among engineering documents, databases, diagrams, and operational systems. By focusing on data organization and enterprise context, organizations can intelligently orchestrate multiple AI models throughout workflows rather than depending on a single model for every task.
Sabet suggests that when evaluating AI platforms, manufacturers should consider data security, intelligent model matching to workflow stages, and measurable improvements in productivity outcomes. Ultimately, the future of enterprise AI lies in intelligently orchestrating enterprise data, semantic understanding, and specialized AI models into reliable workflows that organizations can trust.
Written by urgent.news from GEN Biotechnology's reporting — not their text. Machine-written — it may contain errors, so check the original before relying on it.
