The best optical atomic clocks provide only a correct average frequency. That means that if you have such an optical atomic clock you know with an extremely small uncertainty which is the frequency averaged over a long time interval, of days or weeks, but you know much less about the value of the frequency averaged over a short time, e.g. a second.

The instantaneous frequency and the frequency averaged over short time intervals are not determined by atomic properties, but they are determined by the length of a resonant cavity. That length varies due to vibrations from the environment and due to temperature fluctuations. To minimize these length variations great care is taken for damping vibrations and for stabilizing the temperature, usually in a cryostat, but the accuracy required for an atomic clock is so great that even extremely small residual vibrations and temperature variations limit the performance of the atomic clock.

TFA describes a way to make a laser whose frequency is determined by the atoms whose stimulated emission is used, not by the resonant cavity where they are placed. A resonant cavity is always needed in a laser, to provide the positive feedback that is required for converting an amplifier by stimulated emission into an oscillator that provides a continuous output signal, without an input.

Because the frequency of such a laser is determined by atomic properties, e.g. of barium atoms in the example given in TFA, it is no longer necessary to have a system that tunes the resonance frequency of the laser cavity by measuring atomic resonances in some separate absorption cells, where such a tuning system can only ensure a correct value for the averaged laser frequency.

Nevertheless, this paragraph from the phys.org article is misleading: "Following early theoretical ideas emerged in the 1990s, the concept didn't gain concrete traction until 2008, when researchers at the University of Colorado proposed that superradiant lasers could serve as a new kind of atomic clock.".

This paragraph implies that Reilly at al. from the University of Colorado were the only ones pursuing this idea.

First of all, making this kind of laser has always been a theoretical goal for anyone thinking how to improve atomic clocks, but nobody has succeeded to make a good one. For now, even the authors of this article do not report any conclusive experimental results, so it remains to be seen how this idea will work in practice.

Besides this research of Reilly et al., there have been several research articles published both before 2008 and after 2008, most of them by Chinese authors. The most recent of them, from a few years ago, proposed a plausible way to make such a laser using rubidium atoms or cesium atoms.

However, the main problem with such lasers is the difficulty of obtaining an output signal that is powerful enough to have an acceptable signal-to-noise ratio. IIRC, the Chinese proposal was less likely to ensure a signal-to-noise ratio as good as what seems achievable by the technique proposed by Reilly & al. in the parent article.

So what is described in the parent article seems more likely to be successful than the earlier proposals, but it certainly is not unique.