Outbursts from X-ray binaries (XRBs) are unpredictable events, spanning the broadband spectrum from radio to X-rays. It is believed that outbursts are regulated by thermal-viscous instabilities in the accretion disk (the disk instability model - DIM), but the physical processes that regulate the onset and the evolution of outbursts are far from being fully understood. The long-term behaviour of XRBs during quiescence can give valuable information to test the DIM, but such studies are rare, and in many cases have only been performed at optical wavelengths. One of the best ways to gain a better understanding of the accretion process in XRBs is to combine observations at different wavelengths, in particular X-ray with optical at the very early stages of outbursts. We have been monitoring ~50 XRBs with the Faulkes Telescopes / Las Cumbres Observatory (LCO) global robotic network for more than a decade. We introduce our new real-time optical monitoring pipeline, the "X-ray Binary New Early Warning System (XB-NEWS)", which aims to detect and announce new X-ray binary outbursts within a day of first optical detection. The early optical detection allows us to trigger X-ray and multi-wavelength campaigns at the start of outbursts, which helps constraining the triggering mechanism. This approach was particularly effective for the 2019 outburst of the neutron star (NS) X-ray binary SAX J1808.4-3658, for which an X-ray to optical delay of 4 days was measured with the XB-NEWS pipeline and the X-ray satellites NICER and Swift. These observations provide important insights into the X-ray to optical delay during the onset of outbursts in LMXBs, and therefore to the disk instability model that regulates it (Goodwin et al. 2020). In addition, we present a recent result obtained for the NS-XRB Centaurus X-4. Thanks to observations of the long-term quiescent behaviour of the source, we could predict the potential onset of an outburst several months in advance. Following the first detection of flux enhancement in the optical, NICER and Swift were triggered. The observed activity however resulted in a very faint, short outburst, and we found that the temperature for hydrogen ionization in the disk could not be reached, possibly due to a stall in the propagation of the heating front in the disk. We therefore term this a “misfired outburst”. We predict that faint outbursts like this could be present in more quiescent LMXBs, and only long- term optical monitoring can easily help uncovering them.