One of the themes that emerged repeatedly during RevOps AF Europe 2026 was the challenge of turning AI from an interesting technology into something that creates measurable business value. While many sessions focused on AI strategy, governance, and experimentation, I found myself coming back to a much simpler conclusion: the most valuable AI use cases today are often built on top of data that companies already generate every day.
For RevOps teams, that data is frequently found in sales calls.
One of the most common problems I encounter in CRM projects is incomplete activity data. Sales representatives have conversations with prospects, discuss requirements, uncover pain points, identify stakeholders, and agree on next steps. Yet much of that information never makes it into the CRM in a structured way. Qualification fields remain blank, stakeholders are missing from Opportunities, customer handoffs are inconsistent, and forecasting relies heavily on individual memory.
This is where call recording and transcription technology becomes particularly interesting. As long as calls are consistently recorded and transcribed, AI suddenly has access to a rich source of information that can be used to improve CRM data quality.
Several examples discussed at the conference centered around this idea. AI can review transcripts and identify contacts who participated in a call. It can suggest updates to qualification frameworks such as MEDDIC. It can summarize next steps, identify competitors mentioned during conversations, and even help create Opportunities based on discussions taking place during discovery calls.
Perhaps most importantly, these same transcripts can support customer handoffs. Instead of relying on a salesperson to manually summarize months of conversations, organizations can use AI to review historical interactions and generate a concise overview of the customer’s goals, challenges, stakeholders, and commitments. That information can then be passed to implementation, customer success, or account management teams.
Another concept that appeared repeatedly throughout the conference was the idea of “human in the loop.” Organizations remain far more comfortable with AI making recommendations than with AI making decisions. I believe that is the right approach for most companies today.
For example, rather than allowing AI to automatically assign leads to sales representatives, AI can identify leads that resemble successful customers and suggest them for review. Reps can choose whether to pursue those opportunities, and if they reject them, they can provide a reason. That feedback can be used to improve future recommendations while maintaining accountability and trust in the process.
This recommendation model is particularly attractive because it creates measurable feedback loops. Companies can compare the performance of AI-suggested leads against other leads in the CRM and determine whether the recommendations are actually generating value. The result is a controlled experiment rather than a blind leap of faith.
My biggest takeaway from the conference is that the most practical AI implementations today are not fully autonomous systems. Instead, they are systems that help people make better decisions while reducing administrative work. For many RevOps teams, the path to AI may not begin with complex agents or autonomous workflows. It may begin with something much simpler: making sure every sales conversation is captured, transcribed, and available for analysis.
