Why AI Agents Still Need Digital Adoption
17m
Digital adoption has always been about helping people use the software their company pays for. Now AI is changing what that means. Alan Shimel talks with Tal Carmi, CIO of WalkMe, an SAP company, about where chatbots and AI agents fall short and where WalkMe fits in.
Carmi started as a Salesforce professional and spent years in business applications. He stepped into the CIO role about six months ago, after serving as deputy CIO over information systems, integration, BI and data.
From guidance to cross-application workflows
WalkMe began by guiding users through enterprise applications. Large companies may have thousands of people in a single role, so training and enablement can be harder than building the system itself. Over time, WalkMe added analytics on where people get stuck. It also added simple automations and workflows that start in a CRM and finish in an ERP.
Why chatbots can't replace digital adoption
A chatbot can explain how to use a common app. However, it has never seen a company's own fields, logic and approval steps. Every enterprise system has been customized, Carmi says. As a result, an LLM can only solve a small part of the problem.
AI agents need governance, not just APIs
AI agents face a similar gap. Many enterprise systems still lack APIs, and much of the guidance lives in the user interface. WalkMe works inside the application as the user, with the same permissions and an audit trail. Companies can also decide whether a human must approve every step or only the final one.
Digital adoption for AI itself
Finally, Carmi notes that rolling out an LLM is itself an adoption challenge. People will not use it the same way unless someone shows them how. That makes digital adoption just as important for AI tools as for any other system.