Successful communication is conventionally viewed as a problem of signal-transmission. In fact, all major tech companies (Apple, Google, Microsoft, Amazon, etc) bet heavily on this notion, investing billions. Together with Arjen Stolk and Ivan Toni we wrote a review paper in Trends in Cognitive Sciences setting out why this view is flawed. We discuss theoretical advances in linguistics and cognitive sciences that put the intended concepts, rather than the uttered message, centre stage. This notion isn’t novel, it’s been around at least 100 years, but here we put it to the test and discuss empirical evidence supporting these theoretical frameworks. In a similar fashion, the realisation that communication is solved together, rather than alone, shouldn’t be groundbreaking. But embarrassingly, attempts to reconcile theories of communication with neuroscience have all too often ignored these fundamental aspects of human social cognition. We discuss how instead recent evidence highlights the importance of slowly developing conceptual representations that are shared across communicators. We put forth tentative suggestions how sustained conceptual representations could influence fast paced neuronal processing through dendritic gating. At the moment these are mostly suggestions; we’re not there yet. Currently, the most advanced artificial agents still struggle to sustain a dialogue, and instead process instructions in isolation, or fall back to pre-programmed tropes. I hope this will change soon.