SAP and Infosys executives described principles for scaling AI agents in the enterprise
In the Business Lab podcast (MIT Technology Review) in collaboration with Infosys, representatives of SAP and Infosys explained why moving from pilot AI projects to deployed agents requires a unified data foundation, governance, and evaluating use cases by impact, feasibility, and scalability.
In an episode of the Business Lab podcast, published by MIT Technology Review in collaboration with Infosys, Thorsten Leiduck (SVP, SAP Business AI Platform Ecosystem) and Satinder Pal Singh (associate VP, SAP AI Strategy at Infosys) described, in their words, the key conditions for enterprises to move from isolated AI pilots to integrated autonomous workflows. According to Leiduck, AI is entirely dependent on the data foundation – companies need a connected, semantically rich data layer reflecting their business before they can reliably deploy AI across processes. In applications such as supply chain or financial close, he says, 80% accuracy, which may be sufficient for routine tasks, is not enough.
Singh recommends evaluating the deployment of AI agents ("digital co-workers") by three criteria: impact on business metrics, feasibility (availability of data and talent), and scalability across the organization. According to him, "a pilot that can't be scaled is just an experiment." Both speakers cite a BCG study according to which 74% of organizations remain stuck in the pilot phase without a plan to scale, and an Anthropic survey according to which 42% of companies run into problems with data access and data quality.
Singh further emphasized the role of governance – according to him, the absence of governance does not speed up AI adoption but creates a "risk backlog," which can lead to the need for a disruptive stop and audit. Both see the future in agentic systems coordinating work across applications and in a shift of AI from a productivity tool to the role of a decision-making partner for company leadership.
The text is based on a transcript of the podcast, which is paid content created in collaboration with Infosys; further details of the interview are not provided in the available source.
Why it matters
For companies planning to deploy AI agents, this provides concrete criteria for distinguishing a use case that can be scaled from a mere experiment, and for setting up governance so that it doesn't hinder adoption but reduces the risk of errors in critical processes. For individuals involved in preparing or evaluating AI projects, it offers a framework (impact, feasibility, scalability) that can be used when advocating for or assessing a specific proposal.
Two audiences, two different impacts
What this means
For individuals
Anyone designing or evaluating an AI project at a company can, based on the podcast's recommendations, assess the use case by three criteria – impact on metrics, feasibility with respect to data and talent, and scalability – before investing in further development of the pilot.
For a business
Companies that want to move AI agents from pilots into mission-critical processes (e.g., supply chain, financial close) need, according to SAP and Infosys, a unified data foundation and established governance; without it, uncontrolled risk grows instead of faster adoption.
StrategyCheck the original
Event sources
only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.