For over 20 years it has been clear that accretion onto black holes is ultimately enabled by a magnetic dynamo (MRI) which transports angular momentum outwards, allowing material to accrete inwards. This holds out the possibility to simulate the black hole accretion flow and the associated jet ab initio. However, despite heroic efforts with global (radiative) magnetohydrodynamic simulations in full general relativity, these simulations cannot yet capture the complex observed phenomenology from black hole binaries including discs, X-ray coronae, jets and winds.
Instead, I will show how the data suggest a way forward, with the fast X-ray variability and the jet being directly linked to the MRI, giving testable predictions, while the disc and winds are more dependent on gas and radiation physics.