IBM Consulting says it now runs 6,000 AI agents in 150+ workflows, with forward deployed units taking the model to clients

In an interview with The Financial Express published October 18, IBM Consulting senior vice president Mohamad Ali said the firm has deployed about 6,000 unique AI agents across more than 150 of its own workflows. Its IBM Consulting Advantage platform has an R&D team of over 1,000 engineers, more than 2,000 live projects, and over 500 client deployments that combine digital and human labour.

IBM is taking the same approach to clients through forward deployed units that bring together architects, domain specialists and AI engineers. These teams identify the problem, redesign the workflow, then build and deploy the solution in areas such as finance, HR and banking, and they can work with models and infrastructure from different providers. Ali gave security operations as an example: an investigation that took 45 minutes now takes agents about two minutes, plus ten minutes of human review.

The detail FDEs should note is how IBM measures results. Its Productivity Measurement Lab in Bangalore tests each workflow with and without agents before it goes into production, and Ali admitted that some agent deployments produce negative productivity. IBM says success is judged by productivity and financial outcomes, not by the number of agents or tokens consumed. Proving before-and-after value is becoming part of the deployment job itself.