As Setti & Woltjer noted back in 1973, quasars could be used to construct the Hubble diagram but the actualapplication was not that straightforward. It took years to implement the idea successfully. Most of the ways to use quasars for cosmology now require an advanced understanding of their structure, step by step. We briefly review this progress, with unavoidable personal bias, and concentrate on bright unobscured sources. We will mention the problem of the gas flow character close to the innermost stable circular orbit close to the black hole, discussed 50 years ago, which later led to the development of the slim disk scenario, but was recently revived in the context of Magnetically Arrested Disks (MAD) and Standard and Normal Evolution (SANE) disk models. We also discuss the hot/warm corona issue, which is still under discussion and complicates the analysis of the X-ray reflection. We present the scenario of the formation of the low ionization part of the Broad Line Region as a mostly failed wind powered by radiation pressure acting on dust (FRADO - Failed Radiatively Driven Dusty Outflow model). Next, we discuss the cosmological constraints that are currently achievable with quasars, mostly concentrating on light echo methods (continuum time delays and spectral line time delays with respect to the continuum) which are (or should be) incorporating the progress mentioned above. Finally, we briefly mention future prospects in this direction.